UNIVERSITY OF 
 
 ILLINOIS LIBRARY 
 
 '.7 UR3ANA-CHAMPAIGN 
 
 ENGINEERING 
 
NOTICE: Return or renew all Library Materials! The Minimum Fee for 
 each Lost Book is $50.00. 
 
 The person charging this material isM^eljJMftible for 
 its return to the library from which it was withdrawn 
 on or before the Latest Date stamped below. 
 
 Theft, mutilation, and underlining of books are reasons for discipli- 
 nary action and may result in dismissal from the University. 
 To renew call Telephone Center, 333-8400 
 
 UNIVERSITY OF ILLINOIS LIBRARY AT URBANA-CHAMPAIGN 
 
ENGINEERING LISRARV 
 UNIVERSITY OF ILLINOIS 
 
 URBANA, ILLINOIS 
 
 enter for Advanced Computation 
 
 UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN 
 
 URBANA. ILLINOIS 61801 
 
 01 
 
 M-7i 
 
 CAC Document No. 213 
 
 RECOMMENI TI^N OF COMPUTER SYSTEMS FOR 
 OPERATION OF THE HABITABILITY DATA BASE 
 
 Karl C. Kelley 
 
 and 
 James A. Gast 
 
 September 30, 19?6 
 
SEP 13.1977 
 
 The person charging this material is re- 
 sponsible for its return to the library from 
 which it was withdrawn on or before the 
 Latest Date stamped below. 
 
 Theft, mutilation, and underlining of books 
 are reasons for disciplinary action and may 
 result in dismissal from the University. 
 
 UNIVERSITY OF ILLINOIS LIBRARY AT URBANA-CHAMPAIGN 
 
 i 
 
 L161 — O-1096 
 
CAC DOCUMFNT NO. 213 
 
 RECOMMENDATION OF COMPUTER SYSTEMS 
 FOR OPERATION OF THE HAB1TAB1LIT* DATA BASE 
 
 Prepared tor the 
 
 U . S . Army 
 Construction Engineering Research Laboratory 
 
 under Contract 
 DACAt8-76-M-0?91 
 
 by 
 
 Karl C. Kelley 
 
 and 
 James A. Gast 
 
 Center for Advanced Com rotation 
 University of Illinois 
 Urbana* Illinois 61S01 
 
 September 3 0/ 19^6 
 
CERL HDB Recommendations 
 
 Table of Contents 
 
 1. Summary of Recommendations 1 
 
 1.1 Short Term Recommendations 1 
 
 1.2 Long Term Recommendations 1 
 
 2 . Foreword.. 2 
 
 3 . Approach... 3 
 
 4 . Needs of the Habitability Data B a s p b 
 
 4.1 Conceptual Needs of the Habitability Data Base 6 
 
 4.2 Current Needs of the Habitability Data Base 10 
 
 5. <8S Survey of Interactive Information Systems 13 
 
 6. The Choice of an Appropriate HDR System 23 
 
 6.1 The Choice of Natural Language query 23 
 
 6.2 Suitability of S^ART 31 
 
 6.3 The Choice of Continuino With SMART 33 
 
 7. Developing a Replacement for S f* A <? T 3 8 
 
 7.1 Revision of CELDS for the HDB 39 
 
 1.7. Using a Commercial Information Retrieval System 41 
 
 LIST OF REFERENCES ...47 
 
 BIBLIOGRAPHY 
 
Digitized by the Internet Archive 
 
 in 2012 with funding from 
 
 University of Illinois Urbana-Champaign 
 
 http://archive.org/details/recommendationofOOkell 
 
RECOMMENDATION OF COMPUTER SYSTEMS 
 FOR GPFRATIOM OF THE HABITABILITY DATA BASE 
 
 1. Summary of Recommendations 
 
 The results of this invest ination suoaest a number of 
 courses of action w h i c h will be in the best interests of the 
 Construction Engineering Research Laboratory and their 
 implementation of the Habitability Data Base. We divide the 
 recommendations into short and lon;j term based on the amount 
 of time and effort which will be involved to achieve a 
 measure of satisfaction of the goals of the HDB effort. 
 
 1.1 Short Term Reccmmendat i on s 
 
 a) Reactivate the HDB implementation at the 
 Computer Services Office of the 
 University of Illinois. This holding 
 action will make somethino available 
 until mid-1977* while work is underway 
 to provide a lonqer-tern solution. 
 
 b) Implement the HDB files and versions of 
 the AND/ OR* DOCAX, and BIbAX programs 
 on the Michigan Time Sharing system. 
 
 c) Implement SMART on MTS/ but do the rest 
 of the tasks on the UNIX system at CAC. 
 
 1.2 Lona Term Recommendations 
 
 a) Expand the CELDS work at CAC into a form 
 suitable for both the Environmental IAC 
 and the Habitability IAC. Fool efforts 
 with the CELDS group to acquire 
 appropriate hardware to be used for both 
 sy s t ems . 
 
 b) Convert t h p HDb to a forn» suitable for 
 the Lockheed "DIALOG" system and use it 
 remotely throuqh a Telenet port in 
 Chi cago . 
 
CFRL H D 8 Recommendations 
 
 2 . Foreword 
 
 The recommendations of this report are based on a short term 
 investigation which precluded hands-on trial of most of the 
 systems. The reader is cautioned not to base a lona-term 
 expensive information system implementation solely on the results 
 of examining the features of systems which dre available. A more 
 advisable course of action would be to take steps to contact 
 vendors of the suqgested commercial systems directly to discuss 
 the suitability of their system for the HDP. It is also 
 advisable to use their system on existing data bases to oain a 
 feeling for the kind of response times typically available* the 
 ease with which a search can be made* and the convenience of 
 gaining access to the system. 
 
 It is not possible to determine the real cost/ speed/- or 
 reliability of a system just by looking at the price sheets and 
 the vendor's description of the system. Surveys and literature 
 arches can only guide one in deciding which systems deserve a 
 closer look. They should never be used in the place of hands on 
 experience or a carefully documented benchmark to determine the 
 choice of a system which will be used over an extended period of 
 t ime. 
 
 No claim is made that alt possible systems have been 
 included in our considerations. Work currently underway at 
 various research centers may not be in the current literature. 
 It is suggested that CERL may want to avail themselves of some of 
 
CERL HDB Recommendations 3 
 
 the automated information systems accessible from local sources 
 to make their own search of the literature to obtain 
 bibliographies relating to automated information retrieval. Some 
 of the sources listed in the bibliography of this report 
 themselves contain extensive bibliographies which probably should 
 be examined in areater detail. 
 
 3. Approach 
 
 The search for an appropriate recommendation included a 
 literature survey and an examination of locally known systems 
 which have the potential to meet the needs of the HOP. The 
 literature survey included data base management systems and 
 large-scale interactive information systems. The similarity of 
 the HDB project to another project currently under investigation 
 at the Center for Advanced Computation led to a more thorough 
 comparison of the needs and features of the two systems to 
 determine whether a recommendation to combine the two efforts was 
 w arrant ed. 
 
 The initial literature survey was directed at data base 
 management systems/ since they would "form the backbone of any 
 information retrieval system which involved managino a new set of 
 data (the HD ! J). The essential needs of the H 3 are a data base 
 management system* some type of interactive editing capability* 
 and the ability to recreate the interactive p r o g r a .« b which 
 provide the ^ N D and OR functions available in the earlier version 
 of the HDB. The ability to no the A N D R functions are taken as 
 
CERL HD3 Recommendations 
 
 given in all systems considered- This task is simply not 
 sufficiently complex to be a meaningful consideration in any 
 recommendat ion . Furthermore/ since CfcRL is usually restricted to 
 using time sharing services* it was thought that the choices of 
 computer hardware were restricted to an ILM 360/370 system* a 
 large CDC system known to be available to C E R L * or possibly one 
 of the larger DEC-10 systems* none of which is reknowned for its 
 built-in capability to manage files effectively. For any of 
 these* a sophisticated data base management system would be 
 desirable as a prerequisite to imolementing an HDB. The 
 requirement for managing a large data base was thus taken as a 
 controlling factor and led to the consideration of data base 
 management systems. 
 
 Further experience with the existing HDB documentation and 
 discussions with the technical monitor and programmers involved 
 in the implementation of the S N lART system reinforced the notion 
 that the HDB is not conceptually different from a bibtioaraphic 
 system* either in the way that data is stored or in the kind of 
 programs which would be required to respond to the customer's 
 query. The HDb is essentially a collection of statements* each 
 having been indexed in a special way* and each referring by some 
 means to the original document from which it was taken. This is 
 conceptually similar to a collection of abstracts* indexed by key 
 words or other searchable fields* and each pointing or ?eierring 
 to the original bibliographic citation. Section 4 discusses the 
 
CERL HOB Recommendations 5 
 
 needs of the H D L< as an information retrieval system. 
 
 This conceptual similarity widened t h t Literature se:-rch to 
 include a closer look at the v» e r y specialized t y \. e of i a t a base 
 management systems with query I a n q u a . ; e s which can t e called 
 bibliographic svstems or interactive information systems. 
 Through a series of literature sources we were led to a survey 
 done by the National bureau of Standards in 19 73 (published in 
 1 9 7 A ) . This source constitjtes a reference to the technical 
 features and operational status of interactive information 
 systems/' that is / those providino a 'conversational" u s a o e mode 
 to a "non-pro i? rammer' through a data terminal. In addition to 
 technical information about some A 6 systems/ it provides qui dance 
 in the use of the index to narrow the field of choices in 
 selecting an interactive information system for a particular 
 application. Section b discusses the approach suggested by this 
 reference and i o e s t h r o first . r t i o n of systems 
 
 i ch meet the needs •• f 1 l ■ ■ 
 
 S e c t i on 6 i i s c u s i p r op r i • ■ c >>> AR T a s c choice 
 
 or i m p I e rr en t a t i ' • , z w e II as the implications of 
 
 continuing with S :v ART for I next phase of HOd development. 
 
 ection 7 discusses some of the issues inherent in abandonina the 
 S^IART program and rep I a cinq it with some other s/stem/ whether 
 developed anew or adapted from existing systems. 
 
 A. Meeds of the Habitability Data base 
 
 We seoarate the notions of conceptual needs of the H D t> from 
 
6 CERL HDb Recommendations 
 
 the current needs in the following way: Conceptual needs are 
 based on the problem itself* that is* the problem of storing the 
 HDB and retrievinq the information stored therein on the basis of 
 user requests. Conceptual needs reflect the end user of the HDb. 
 Current needs* on the other hand* deal with the more practical 
 immediate concerns of the HDB effort* making the service 
 available to the current set of users in a cost effective manner 
 as quickly as possible. The current need seems to be primarily 
 for a system which will run the A N D R programs and allow StfART 
 requests to be submitted in a batch mode. We include in current 
 needs any system which could do the user's end function as 
 effectively as SPiART* without extensive reprogramming or 
 reformatting of the data base. 
 
 4.1 Conceptual Needs of the Habitability Data base 
 
 Conceptually* the hDR consists of a set of statements drawn 
 from an appropriate literature. These statements are formulated 
 by trained specialists who not only condense the information in 
 the literature* but also classify the information by indexing the 
 statement. This process is conceptually equivalent to abstracting 
 a document and providing an index classification of the document. 
 Whereas most bibliographic retrieval systems keep the information 
 about the document in the same record as the abstract (and 
 possibly key words in addition to author* title* etc)* in He HDB 
 the only information directly linking the entry in HDb with the 
 original source of the information is a document number encoded 
 
CERL HDB Recommendations 7 
 
 as part of the sequence number field of the HDb statements. The 
 index of the document is a multidigit strinq of codes which is 
 prepended to the first card image of a particular statement. 
 
 The user of the HDb wishes to formulate a simple request to 
 retrieve information which is of immediate concern to him. With 
 the HDB as originally designed* this request is stated in terms 
 of the classification of the statement as represented by the 
 index. This index is comprised of 10 coded values: Lbl 
 
 F U N C . . . a three digit functional area code 
 
 TRFC...a "> digit training facility code 
 
 PHYS...3 1 digit physical setting code 
 
 AENV...a 2 digit "A" environmental descriptor 
 
 BENV...a 2 digit M F" environment descriptor 
 
 0CCU...a 1 digit occupant code 
 
 P S T R . . . a 1 digit coie for posture of people 
 
 I N V M ... a 1 dioit code for involvement of people 
 
 0R6F...S 1 digit code for organizational functions 
 
 SFCN...a 1 digit code for function of the state merit 
 
 A more complete description of the classification and 
 indexing scheme is given in C 1 J . 
 
 The primary programs for selecting statements interactively 
 on the basis of the indexes are the A'mD and OR oroarams which run 
 interactively on the D E € - 1 U system as part o * the Prototype HDB 
 C43f53C31. These programs use a rather forced dialog to input 
 the appropriate fields which are to be searched on and the values 
 to be searched for. The interactive response is a set of 
 statements/- along with the appropriate document number and the 
 number of this statement with respect to the source occument. 
 There is no capability tc net a count of documents which meet one 
 
CFRL H D 8 Recommendations 
 
 criteria or set of criteria and then deter mininq whether that set 
 should be farther limited by ANDinq with another set. The entire 
 request is made at the outset/ and the entire set of documents 
 which match the request is printed as output. There seems to be 
 no capability of savinq the numbers of these documents for later 
 refinement by further search requests. 
 
 Two other programs exist in the Prototype HDP system which 
 reflect both conceptual and current needs. The function of the 
 programs is to allow the user to see the bibliographic citation 
 of a document if he knows the document number and to see the text 
 of the document if he knows the number. Note that AN DOR returns 
 the statement and the number of the document. (The document is 
 not really there/ it is just the collection of all statements 
 which came from that document) 
 
 This technique of finding statements is perhaps approoriste 
 to some potential users of the HDB. In particular/ the person 
 who wishes to write a criteria manual for design of a certain 
 training facility might want to retrieve what is available and 
 related to that kind of facility. How ever/ information 
 specialists responding to a submitted query/ and to some extent 
 the end customer himself/ might find that a better way of 
 expressing the inquiry and conducting the search is needed. The 
 HDB does not contain keywords which can be used to characterize 
 content of statements. (In its present form/ content is only 
 characterized by the index digit string). Thus a retrieval 
 
CERL HDB Recommendations 
 
 system based on full-text search of the statements*' preferably 
 with natural language input/- is a second conceptual need of the 
 HDB. 
 
 At the present time this need is met by the collection of 
 programs known as the SMART system. This system operates in 
 batch mode on the IBM 36 system. It has been implementeo 3 1 the 
 Computer Services Office (University of Illinois at Urban a) as 
 part of the prototype HDR effort. "The system takes documents and 
 search requests in English/ performs a fully automatic content 
 analysis of the texts/ matches analyzed documents with analyzed 
 search requests/ and retrieves those stored items believed to be 
 most similar to the queries. Among the language analysis 
 procedures incorporated into the system are word suffix cutoff 
 methods/ thesaurus lookup procedures/ phrase generation methods/ 
 statistical term associations/ syntactic analysis/ hierarchical 
 term expansion/ and others. " C 6 3 
 
 As a part of the SMART user interface for the prototype HDB/ 
 a prooram on the DfcC-10 computer accepts input o'f Query 
 submittals and formulates batch jobs for the 360. These jobs are 
 submitted across a link to the batch machine. The user returns 
 later to see if his job is done and relieves his output 
 (responses to his query) by runninq another prooram on the 
 DEC-10. The time lao between request and response has not been 
 satisfactory with the present implementation. 
 
10 
 
 CLRL HDB Recommendations 
 
 What is needed and is missing in the current implementation 
 is an interactive on-line version of SMART. Salton recognized 
 this as a need C63. To run SMART interactively would require a 
 different operating system on the 360* namely one that allows for 
 time-shared user interactive terminals. There have been no major 
 updates of SMART since the library was obtained from Cornell for 
 the prototype HDB. At latest report* no interactive version of 
 SMART is available in release form* although some effort was 
 expended at Cornell in implementing an interactive version under 
 the IBM TSO operating system. Even if that were successful* it 
 would be of little value to any solution which proposes using the 
 360 at CSG* since that system will stay batch until its eventual 
 ret i r ement . 
 
 If the conceptual need for natural lanquaae processino of a 
 query is artificial* some of the systems to oe mentioned in 
 Section 5 would probably well serve the needs of the HDB. 
 
 4.2 Current Needs of the Habitability Data Base 
 The current need of the HDP is a system which provides for 
 the conceptual needs outlined above as well as the more immediate 
 concerns of finding an appropriate operating system and computer 
 to run it on. The scope of work for this contract lists five 
 definitions of the needs of the HDB. Two of these fall into the 
 class of con rtual needs: 
 
CERL HDB Recommendations 
 
 11 
 
 1) the types of programs currently in use must be 
 ava i table 
 
 ?) the system has the capability of handling summary 
 data as well as bibliographic and textual data 
 
 These have been examined in the section on conceptual needs. 
 The other needs are current needs discussed in this section. 
 
 The text -editing capability is desirable so that 
 corrections and changes can be made to the HDB statements/ and so 
 that new statements can be added as the collection nrows. Any 
 system which will be capable of the interactive access required 
 for the AND/OR programs will/ without exception/ have text- 
 e d i cinq capability. So long as the HDB statements are part of a 
 non-specific text file/ they be accessible and editaole with the 
 editors on most systems. 
 
 However/ the capability to edit HD r - statements which are 
 already included in a data base which has under a one sotiip o'earee 
 of inversion might be somewhat of a problem. The typical 
 retrieval system requires that the data and the fields which will 
 be searched be made ready for a larqe inversion process which is 
 run against the data base to get it properly organized for faster 
 retrieval. In some organizations this data base inversion 
 process is very time consumino. The capability to access the 
 statements independent of indexes to the statements is thus a 
 requirement for on-line correction to the HDB statements. 
 Similarly/ in order to keep the data base updated/ i* should L>e 
 possible to input new statements in text form. This is not a 
 
12 
 
 CERL HDB Recommendations 
 
 problem. However/ it is quite likely that before new statements 
 can be used as an integral part of the HDB* the inversion process 
 must be run aoain. This would restrict updating to periodic 
 updates of perhaps once a month. This is the norm rather than the 
 exception in data base systems of the capability described. 
 
 Commercially available systems will automatically be aole to 
 take care of control and billing of outside users (they make a 
 living doing it). University computer centers sometimes have 
 more difficulty with this in that their process for establishing 
 user accounts is sometimes rather cumbersome. However/- the 
 systems under consideration and outlinea in the accompanyina 
 recommendations all meet the criterion that outside users can be 
 admitted to the system and billed directly. Similarly* the 
 capability for remote Low-speed access fro^ terminals should be 
 taken as given in all of the systems under discussion here. The 
 only systems for which this is not the case are systems for which 
 access is restricted to remote batch/ and such a system cannot 
 meet the editing and interactive requirements. Where necessary* 
 submission to batch systems should be accomplished via an 
 interactive system* similar to the technioue used between the 
 DEC-10 and the 36(1 in the prototype HDB work. This should 
 always* however* be considered as clumsy snd not conducive to the 
 kind of immediate feedback to be obtained with interactive 
 
CERL HDO Recommendations 
 
 13 
 
 systems such as those commercially available. 
 
 5. NBS Survey of Interactive Information Systems 
 
 The National Bureau of Standards has already anticipated the 
 
 need for government agencies to consider the choice of an 
 
 interactive information system. A report published in 1974 
 
 constitutes a reference to the technical features and one rational 
 
 status of such systems available at the time [?3. From the 
 
 introduction to that report: 
 
 "This report is written for the purpose of 
 providing Federal ADP customers with information 
 on a certain class of computer systems which are 
 capable of handling scientific and technical 
 inform ation„ The report attempts to show what is 
 available and to characterize these systems in 
 such a way as to answer questions which naturally 
 arise prior to selecting such a system for a 
 particular installation. The report is written at 
 a level of technical detail which is aimed at 
 information specialists rather than programminq 
 experts. It is intended to be informative and 
 instructive* and not critical or evaluative." 
 
 "We have reviewed tor inclusion in this index 
 over 200 systems which came to our attention from 
 various published and unpublished sources as well 
 as from word-o f-mou t h . The systems which were 
 selected conform to the following definition: 
 "Information Retrieval" or "Data Management" 
 packaqes or services which are available to any 
 Federal ADP installation* and which offer an 
 interactive query and search capability that is 
 geared tor use by non-programmers." 
 
 They eliminated from consideration systems which: 1) are 
 batch systems* I) have query languages not for use by non- 
 pro q rammers/ .">> are in research or development/ 4) a.e no longer 
 supported/ 5) are no lonqer in business or locatahle/ 6 ) are 
 
14 
 
 CERL HDB Recommendat ions 
 
 subject to legal or security p r o b I e m s in the way of releasing the 
 system/ or 7) were not documented. 
 
 It seems at least strongly suggestive that these systems 
 meet the basic needs of the US Army CERL* if one of them meets 
 the specific conceptual needs of the HDB. 
 
 The intent of this section is to examine the organization of 
 that report and to frame current concerns in terms of the 
 selection criteria outlined therein. Table 1 is a list of the 
 systems which met the criteria for inclusion in this survey. 
 Table 2 is the questionnaire which was usee? to characterize the 
 features of the various systems. Included in the reoort is a 
 summary of the features of each of the examined systems/ listed 
 in a manner similar to the format of the Questionnaire. 
 
 However^ before examining each of the systems reported/ the 
 suggestion is mede that the needs of potential users of the 
 system be classified in order to make a first cut at system 
 selection,, Their recommendation for a first elimination is based 
 upon potential usage and estimated cost first/ then on the 
 availability of a given main-frame/ ar\cj in the case of a 
 requirement for a specific data base/ on the availability of that 
 data base as a service. In the case of the HDB investigation/ 
 several choices of main-frame are available/ and it has not been 
 determined whether a package should be put up on one of these 
 mainframes or a service bureau should be used. Sinc^ a decision 
 can be made on these choices at a later time/ we can proceed 
 
CERL HDB Recommendations 
 
 15 
 
 Name 
 
 BASIS 
 
 CDMS 
 
 CIRCOL 
 
 (Data/ Central) 
 
 DIALOG 
 
 DMARS 
 
 DML 
 
 DRS 
 
 DS/3 
 
 EMESARI 
 
 ENFORM 
 
 FLEXIMIS 
 
 GIM 
 
 GIPSY 
 
 IMARS 
 
 IMS(OEP) 
 
 IMS/360 
 
 IMS/8 
 
 INQUIRE 
 
 INSYTE 
 
 LEADERMART 
 
 MARK IV 
 
 MARS III 
 
 Name 
 MARS VI 
 
 MASTER CONTROL 
 MICROTEXT 
 MINIDATA 
 MIRADS 
 MUSE 
 NASIS 
 N.Y. TIMES 
 OLIVER 
 ORBIT III 
 PIRETS 
 
 QUERY UPDATE 
 RAMTS 
 RECON 
 RFI 
 RIQS 
 SHOEBOX 
 SOLAR 
 SPIRES II 
 STAIRS 
 SYSTEM 2000 
 TICON 
 UNIDATA 
 
 TABLE 1. SYSTEMS INCLUDED IN THE NBS SURVEY 
 
16 
 
 Ct-FL HDB Recommendations 
 
 o 
 
 CI 
 
 
 
 
 
 !"•• 
 
 
 
 » 
 
 -3 -i 
 
 
 
 •k 
 
 
 4-> 
 
 
 >i 
 
 • 
 
 C ri 
 
 
 
 ^ 
 
 
 o 
 
 
 M 
 
 C7i 
 
 K <*H 
 
 
 
 
 
 id 
 »-i 
 
 
 rH 
 Id 
 
 01 
 
 o -c 
 
 
 
 c 
 
 
 tu 
 
 
 U 
 
 
 c c 
 
 
 
 c 
 
 
 H! 
 
 
 ■H 
 
 V 
 
 •rl nj 
 
 
 
 .— 1 
 
 Id J 
 
 id 
 
 4-1 
 
 x> 
 
 in 
 
 
 
 
 
 id 
 
 
 C 
 
 >< 
 
 4! nj 
 
 
 
 >i 
 
 to 
 
 
 C 
 
 
 
 
 
 •D 4J 
 
 
 to 
 
 c 
 
 id 
 
 u 
 
 > 
 
 •3 
 
 
 
 id 
 
 
 d3 
 
 It 
 
 
 •H 
 
 id 
 
 •H 
 
 a tv 
 
 14 "3 
 
 
 M 
 
 
 X 
 
 r; 
 
 xj 
 
 
 in U) 
 
 o 
 
 
 6 
 
 14 1 
 
 u- 
 
 a 
 
 
 u^ 
 
 u 
 
 Si 
 
 
 o 
 
 C 
 
 -J 
 
 id 
 
 
 
 
 
 id o 
 
 
 
 
 
 c 
 
 r< 
 
 •p 
 
 
 j: 
 
 c 
 
 
 u 
 
 u 
 
 
 tr. 
 
 
 u 
 
 c 4J 
 
 o 
 
 
 
 X 
 
 V) 
 
 C 
 
 o 
 
 % 
 
 •H 3 
 
 4J IT, 
 
 
 tH 
 
 
 
 Id 
 
 ■rl 
 
 rH 
 
 id 
 
 C -H 
 
 
 (0 
 
 u 
 
 
 rH 
 
 JU 
 
 L 
 
 cc 
 
 E. -C 
 
 «"•■ 
 
 3 
 
 
 4J 
 
 ^J 
 
 •rl 
 
 3 
 
 •o u 
 
 t 
 
 10 
 
 ~ 
 
 V 
 
 Id 
 
 •rl 
 
 (0 
 
 i- 
 
 •H C 
 
 ii 4-1 
 
 o 
 
 •H 
 
 u 
 
 id 
 
 „T2 
 
 01 
 
 
 o 
 
 c 
 
 u 
 
 > 
 
 rj 
 
 a 
 
 
 C 
 
 
 
 •rl W 
 
 O 
 
 a 
 
 •rl 
 
 r C 
 
 r-( 
 
 r~H 
 
 CM 
 
 <4. 1 -a 
 
 J= c 
 
 u 
 
 dd 
 
 C 
 
 
 n 
 
 
 -Q 
 
 Vh 
 
 4-1 
 
 o 
 
 c 
 
 E. 
 
 U 
 
 
 
 •P 
 
 Id T3 
 
 rH 
 
 3 
 
 •H 
 
 c 
 
 (3 
 
 •h 
 
 •H 
 
 • rl 
 
 CJ 1 
 
 C C 
 
 u 
 
 
 u 
 
 
 c. 
 
 
 In 
 
 E N 
 
 C E 
 
 — 1 
 
 c 
 
 u 
 
 >H 
 
 >. 
 
 11 
 
 n; 
 
 id 
 
 O -n 
 
 K 
 
 c 
 
 c 
 
 C 
 
 u 
 
 1-1 
 
 > 
 
 c ~ 
 
 41 d H 
 
 
 •u C 
 
 ^ OH 
 
 
 o«o 
 
 O O 1 
 
 
 4H It! 
 
 t. c: 
 
 
 r- 
 
 «H 
 
 C 
 
 >. a 
 
 Vi tj: 
 
 id 
 
 4J q; t4 
 
 t; 
 
 
 ■H u o 
 
 t — 1 41 
 
 3 
 
 riOO 
 
 O 0"i 
 
 4J 
 
 H O 
 
 CO-" 
 
 
 £) -H l-l 
 
 ■h rH jd 
 
 il 
 
 id X3 
 
 4J I. 
 
 Ih 
 
 f-. C CJ 
 
 ic ai 
 
 a 
 
 fld£ 
 
 ■c ■?. - 
 
 u< 
 
 U 4-1 
 
 a, »j -d 
 
 
 
 >, 
 
 3 »h 
 
 u 
 
 rj rH *4.< 
 
 'u 
 
 
 OJ c c 
 
 o o o 
 
 - 
 
 4J > 
 
 c o 
 
 & 
 
 rj -H Ul 
 
 •rtOi, 
 
 i: 
 
 HU 
 
 r-| 0. 
 
 •H 
 
 UUP 
 
 i c: n 
 
 44 
 
 CDC 
 
 c. <z 
 
 •rH 
 
 «H 13 
 
 C -C V-4 
 
 'C 
 
 k a x; 
 
 = U C 
 
 •j 
 
 id U"- o 
 
 c 
 
 O = 
 
 •H 1/1 
 
 O o ^• 
 
 i; c d 
 
 3 4-> 4J 
 
 u-i O id 
 
 Ci 'C 
 
 C O C 
 £U-h 
 
 V = F 
 O 
 
 
 
 V4 x: c 
 
 C T3 
 
 id a CP 
 
 >h t; o 
 
 in -j -h .c 
 
 4) TJ 4J 4J 
 C 14 »i 
 
 O ~H C 
 
 c x -h 
 o >. o 
 
 rH rH I t» 
 
 rj In »-. 
 
 id c c. o 
 
 4-1 O O 
 
 Id -H O 
 
 V 4-) 4-> >H 
 
 C1C4J-3 
 
 c o c 
 
 r( C O IT 
 
 C V. o 
 
 O (.'. o 4J 
 
 c-j 1< c 
 
 •r, t, a r-. 
 
 x: c oi 
 
 C O 4J O rH 
 d CJlH-H 
 
 u« ; 0*4 
 
 '3 
 
 id 
 
 CP 
 
 O 
 
 
 c 
 
 "U t; 
 
 u 
 
 ■H 
 
 o; -h 
 
 :< 
 
 4J 
 
 17 
 
 o 
 
 Id 
 
 4-1 
 
 o r> 
 
 id o 
 
 
 ■H 10 
 
 14 c 
 
 > 
 
 O 
 
 V -ri 
 
 Id 
 
 c w 
 
 C X 
 
 rl 
 
 •n rj 
 
 V4 O 
 
 a it 
 
 C 
 
 ii 
 
 11 c 
 
 m 
 
 r~ o. 
 
 ^ 
 
 
 
 a 
 
 a r( 
 
 i'. 
 
 O -3 
 
 c 
 
 
 C C 
 
 a 
 
 o 
 
 id 
 
 
 
 4-1 
 
 o 
 
 4) -H 
 
 
 (rt 
 
 ri 4-1 
 
 l: 
 
 Cu-3 
 
 £2 
 
 i/ 
 
 C -1 
 
 o 
 
 •H C 
 
 *J c 
 
 o 
 
 C S- 
 
 4J 
 
 c 
 
 > >, 
 
 11 3 
 
 Vh 
 
 O 
 
 ri id 
 
 c. 
 
 ■o .« 
 
 
 w 
 
 =-< 
 
 Q 
 
 z 
 
 ►J 
 
 c 
 
 r4 
 
 C 
 
 S3 
 
 2 
 
 Ul 
 
 r^ 
 
 < 
 
 U 
 
 rH 
 
 £h 
 
 ~; 
 
 ►.3 
 
 fe- 
 
 H 
 
 l-H 
 
 
 
 ^ 
 
 t- 
 
 t! 
 
 Q 
 
 Hi 
 
 l~ 
 
 rJ 
 
 K 
 
 u 
 
 r> 
 
 c 
 
 H 
 
 t: 
 
 
 < 
 
 < 
 
 n 
 
 K 
 
 M 
 
 ^_i 
 
 
 t; 
 
 c 
 
 C-. 
 
 c^ 
 
 K 
 
 < 
 
 !£ 
 
 J 
 
 ^ 
 
 
 C 
 
 VH 
 
 
 
 
 t. 
 
 • 
 
 • 
 
 • 
 
 y 
 
 KH 
 
 < 
 
 J 2 
 
 lz 
 
 1 C 
 
 u 
 
 r. m u 
 
 Lh 
 
 Ct-h 
 
 73 
 
 < < 
 
 t~i 
 
 i; ri u 
 
 <■ 
 
 
 o u zj 
 
 U 
 
 
 -J 
 
 <. id J3 
 
 
 
 
 c 
 
 
 < 
 
 
 
 
 £- 
 
 Ui 
 
 c 
 
 
 < 
 
 x 
 
 z: 
 
 
 C 
 
 u 
 
 rH 
 
 
 
 U3 
 
 K 
 
 
 a 
 
 r^ 
 
 a 
 
 
 ti 
 
 
 
 a <. 
 
 7^ 
 
 j - 
 
 
 C Eh 
 
 o c 
 
 * 
 
 z 
 
 c < 
 
 n r; 
 
 g 
 
 
 
 a 
 
 [. M 
 
 M 
 
 rj 
 
 U E< 
 
 '^ 
 
 Eh 
 
 <C " 
 
 J rH 
 
 c 
 
 < 
 
 r. u; 
 
 tn C 
 
 c 
 
 U 
 
 c e 
 
 .O U3 
 
 c. 
 
 M 
 
 •-< u 
 
 
 u; 
 
 rj 
 
 C' Eh 
 
 
 a 
 
 X 
 
 C- 
 
 j 
 
 1-i 
 
 r 
 
 a 
 
 O 
 
 
 H 
 
 p. 
 
 c; < 
 
 r-< 
 
 M 
 
 1 
 
 
 F< 
 
 U. £-■ 
 
 
 
 
 
 
 
 i 
 
 H 
 
 u 
 
 C. 
 
 ^; 
 
 H 
 
 H 
 
 CJ 
 
 ;: 
 
 L> 
 
 < 
 
 < 
 
 < 
 
 1. 
 
 U 
 
 E 
 
 
 
 
 •4H 0) 
 
 
 
 
 4J XJ 
 
 
 
 
 C K C 
 
 
 
 
 >n Id 
 
 
 14 
 
 in 
 
 > V) 
 
 
 
 
 ■H 
 
 •H W 
 
 
 4-1 
 
 
 cn 
 
 
 3 
 
 Mj 
 
 • a 
 
 
 A 
 
 rH 
 
 c in 
 
 
 C 
 
 U 
 
 lH C rH 
 
 
 
 
 w CJ 
 
 n 
 
 
 
 
 4-1 h 
 
 •rl > 
 
 
 
 ■-H C 
 
 t: w 
 
 
 
 
 4J 
 
 In -3 
 
 
 jn 
 
 Cv W 
 
 £ 
 
 
 4J 
 
 tji 3 
 
 n > w 
 
 
 
 Id O • 
 
 C -H 
 
 
 '4-. 
 
 .* C 
 
 
 
 lH 
 
 -H 
 
 O k O 
 
 ■-< > u 
 
 c 
 
 
 IJ O'H 
 
 Id -rl C 
 
 J-l 
 
 c 
 
 avi 4J 
 
 C 4J 4J 
 
 c 
 
 •H 
 
 >d 
 
 COd 
 
 
 
 
 E "J rH 
 
 •h r; c 
 
 E 
 
 T3 
 
 nj <y rH 
 
 4-1 M -n 
 
 0) 
 
 O 
 
 u v. .rj 
 
 ■ri tJ 1 rH 
 
 rH 
 
 CT' O 4-> 
 
 t! iJ -H 
 
 a 
 
 r-t 
 
 Own 
 
 — rH U 
 
 E 
 
 •H 
 
 l4»4 C 
 
 < id 
 
 •r4 
 
 tH 
 
 a -rt 
 
 '3 
 
 OJ 
 
 0; 
 
 rl 
 
 r( 
 
 O 
 
 l-i 
 
 'M 
 
 •rH 
 
 '1. 
 
 u. 
 
 
 
 O 
 
 41 
 
 1*3 
 
 
 
 O. 
 
 ■H 
 
 
 > 
 
 0) 
 
 rl 
 
 X • 
 
 41 
 
 4-1 "3 
 
 V) 
 
 
 
 
 M 
 
 41 
 
 14 O 
 
 u 
 
 !4h 
 
 
 
 /: vi 
 
 j: 
 
 5 c 
 
 ~s 
 
 c w 
 
 c 
 
 1 ri 
 
 ■rl • 
 
 
 "3 
 
 "J >4H 
 
 •a 
 
 O rH 
 
 >H 
 
 rl 4) 
 
 rH O 
 
 rH 1/1 
 
 r~i 44 
 
 •H 4-1 
 
 •H 14H 
 
 Cm H 
 
 b. 
 
 c 
 
 
 ^d 
 
 
 O 
 
 
 X 
 
 
 O 
 
 
 •r| 
 
 JZ 
 
 X3 
 
 rr id 
 
 i 
 
 (0 3 3 
 
 
 44 04d • 
 
 v. 
 
 >-J 1h 4J 41 
 
 E 
 
 to .c dn 
 
 O 
 
 3 4J - id 
 
 4-1 
 
 O >H 
 
 10 
 
 C >-T3 
 
 >- 
 
 nH04) 
 
 to 
 
 c w » 
 
 
 •H 3 
 
 U) 
 
 4-J c 4J tr- 
 
 
 
 rH O C C 
 
 ■r| 
 
 3 C «-< 
 
 •4J 
 
 F; fl r. > 
 
 ■r| 
 
 4)4I M 
 
 44 
 
 OH C 41 
 
 c 
 
 > 3 <n 
 
 
 
 M E 1 c 
 
 '3 
 
 --I 
 
 M 
 
 O 01 u 
 
 Eh 
 
 
 
 
 >i 
 
 
 
 
 
 ^ 
 
 
 
 z 
 
 
 
 
 
 
 
 
 
 CO 
 
 
 
 
 Eh 
 
 
 X 
 
 
 
 O 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 M 
 
 
 i-< 
 
 
 fcj 
 
 H 
 
 
 
 n 
 
 
 
 
 
 
 
 
 
 F^ 
 
 
 
 
 ►^ 
 
 
 M 
 
 
 « 
 
 Eh 
 
 
 14 
 
 Fi 
 
 
 Eh 
 
 
 
 
 
 10 
 
 
 j~^ 
 
 z 
 
 
 
 w 
 
 
 J 
 
 W Cd 
 
 < 
 
 a 
 
 
 O 
 
 < 
 
 
 z 
 
 
 
 
 
 w 
 
 
 a? 
 
 
 
 
 
 CQ 
 
 
 M 
 
 U H 
 
 ■3; 
 
 i/i 
 
 M 
 
 M 
 
 W 
 
 w 
 
 
 "C 
 
 
 
 Ei 
 
 
 M 
 
 
 
 rH 
 
 1-J E-i Eh 
 
 sa 
 
 m Eh ;: 
 
 £•< 
 
 >-4 
 
 Z 
 
 ^ 
 
 £- 
 
 CJ 
 
 t: 
 
 
 4J 
 
 w 
 
 
 g 
 
 
 
 Eh 
 
 
 
 M 40 4.Q 
 
 5 
 
 > to 
 
 r.-< 
 
 < 
 
 £h O 
 
 « 
 
 H 
 
 < 
 
 z 
 
 
 Eh 
 
 u 
 
 10 
 
 a ui 
 
 
 Ch 
 
 
 
 n 
 
 < c 
 
 K O 40 t-i O) 
 
 
 
 £h 
 
 Z rH 
 
 fcl 
 
 7d 
 
 Ul 
 
 
 
 
 0: 
 
 F^ 
 
 u 
 
 
 c K 
 
 
 M 
 
 
 
 < 
 
 ." tj to u 
 
 M 
 
 u] CJ 4-3 tr. M 
 
 U) 
 
 to 
 
 < t- 
 
 to 
 
 M 
 
 ^J 
 
 « 
 
 
 >< 
 
 
 
 t. 
 
 z 
 
 u u. 
 
 • 
 
 K 
 
 w 
 
 
 ts 
 
 U W3 
 
 < 
 
 40 4^ :j « 
 
 
 ;i 
 
 u < 
 
 H rJ 
 
 
 
 
 M 
 
 
 Ul 
 
 ■': 
 
 >■ 
 
 c 
 
 to 
 
 <AJ 
 
 U 
 
 1 
 
 cd 
 
 < 
 
 V. Eh rC £h 
 
 I> 
 
 Eh < 4J < 
 
 C-. 
 
 M 
 
 4*4 
 
 rj 
 
 B 
 
 ^ 
 
 ;: 
 
 
 '.'. ■ 
 
 £h 
 
 t-H 
 
 E- P 
 
 w 
 
 C w 
 
 
 a < u «•■: 
 
 'C 
 
 X to 
 
 c 
 
 
 [.. rl 
 
 
 
 CJ 
 
 i" 
 
 ^d 
 
 z. 
 
 45 
 
 < 
 
 
 0. 
 
 z 
 
 
 tl) 
 
 
 Eh Z 
 
 W 
 
 0. g J g 
 
 
 
 E- 
 
 H < 
 
 
 rH 
 
 w 
 
 4>1 
 
 s 
 
 Z 
 
 rJ 
 
 k! 
 
 V. 
 
 K-rH 1 
 
 H 
 
 O 
 
 
 < 
 
 K 
 
 w 
 
 1--; S < w «c 
 
 >• 
 
 40 
 
 p; Eh 
 
 >< 
 
 »> 
 
 to 
 
 
 M 
 
 
 < 
 
 VH 
 
 i-l 
 
 
 jr 
 
 
 < 
 
 « « * 
 
 u 
 
 < Eh :< C« 
 
 K 
 
 r: 
 
 U 40 
 
 K 
 
 1^ 
 
 1-4 
 
 a 
 
 C-. 
 
 i< 
 
 tl 
 
 z 
 
 sc 
 
 Eh ft. 
 
 1-4 
 
 td 
 
 M E 
 
 2 
 
 O Eh O E-i 
 
 H 
 
 Ci) Eh < < < 
 
 O 
 
 M 
 
 M Z 
 
 O 
 
 u 
 
 a 
 
 t: 
 
 
 < 
 
 4J 
 
 H 
 
 S3 
 
 Z M 
 
 P3 
 
 g 
 
 Eh 
 
 O 14 
 
 £ 
 
 t, < t . < 
 
 > 
 
 W <; CI Eh Q 
 
 Ch 
 
 b. 
 
 40 M 
 
 t< 
 
 to 
 
 Ch 
 
 £-■ 
 
 7' 
 
 d 
 
 0: 
 
 s 
 
 z 
 
 td £h 
 
 •^ 
 
 to 
 
 M r4 
 
 
 c 
 
 
 W 
 
 
 
 40 
 
 
 
 Z-j 
 
 M 
 
 u 
 
 y 
 
 g 
 
 1 rj 
 
 EH 
 
 w 
 
 >H 
 
 K W 
 
 O 
 
 • • 
 
 W 
 
 • » • 
 
 M 
 
 • 
 
 
 H 
 
 • 
 
 * 
 
 t. 
 
 s 
 
 a. 
 
 c 
 
 K 
 
 K 3 
 
 s 
 
 en 
 
 O £h 
 
 01 
 
 id ja 
 
 dl 
 
 < .0 (J 
 
 S 
 
 id 
 
 £ 
 
 S 
 
 id 
 
 .0 
 
 6 
 
 O 
 
 4/2 
 
 E- 
 
 E-i 
 
 K 5T 
 
 
 C 
 
 • 
 
 
 • 
 
 
 • 
 
 
 • 
 
 
 
 • 
 
 
 
 CJ 
 
 
 
 
 
 
 
CERL HDB Recommendations 
 
 17 
 
 
 
 
 
 >, 
 
 e 
 
 X 
 
 
 
 Si X 
 
 >1 
 
 £. 
 
 
 
 
 
 
 
 
 
 
 
 +J 
 
 
 
 
 
 
 
 X 
 
 CJ 
 
 -u 
 
 
 
 4J 4-1 
 
 ■y 
 
 'J 
 
 
 
 
 <r. C 
 
 
 
 
 x; 
 
 
 a 
 
 n 
 
 
 
 
 u 
 
 
 ■r4 
 
 43 
 
 •H 
 
 
 
 •H -H 
 
 ■H 
 
 M 
 
 
 
 
 
 
 
 
 (J u c 
 
 
 c 
 
 
 <3 
 
 
 
 
 p. 
 
 X 
 
 J 
 
 •> 
 
 
 
 •-I !2 
 
 i-l 
 
 13 
 
 
 
 x: 
 
 o -y 
 
 CJ 
 
 
 
 i-i x; n c 
 
 
 > 4J 
 
 c < v - 
 
 4J 
 
 
 
 c 
 
 XI 
 
 ■* .C 
 
 
 
 
 
 ■h x: 
 
 •H 
 
 O 
 
 
 
 o 
 
 IP 
 
 c 
 
 ■y 
 
 
 IB t" >-i-i 
 
 
 ■H C 
 
 H C 
 
 x: ic 
 
 
 
 c 
 
 O X CJ 
 
 Si 
 
 c 
 
 
 
 £> U 4J 
 
 -U 
 
 t" 
 
 
 
 u 
 
 3 4J 
 
 c 
 
 X 
 
 
 o -h o n 
 
 
 Ci -S 
 
 x: o 
 
 01 ~J 
 
 m 
 
 
 >>-H 
 
 14 
 
 <r u 
 
 c 
 
 c 
 
 
 
 IS 'H -H 
 
 13 
 
 
 
 
 13 
 
 fV 
 
 iA 
 
 c 
 
 
 «1 --4 >-. C -1 
 C" C 3 O O 
 
 
 ^ 
 
 
 ■r4 
 
 >1 
 
 
 04 44 
 
 a 
 
 x -t 
 
 •r4 
 
 •fi 
 
 
 
 a <3 
 
 C- 
 
 Li 
 
 
 
 
 
 C Ji 
 
 
 y 
 
 
 
 x: 
 
 C t: 
 
 — o 
 
 * Si 
 
 
 •h <3 
 
 c 
 
 13 CI 
 
 x o 
 
 C! 
 
 
 
 COt' 
 
 IB 
 
 a 
 
 
 
 CI 
 
 4-i y 
 
 j; 
 
 
 
 in C *-4 r: V4 r: 
 
 
 o c 
 
 4-1 f. 
 
 S x: 
 
 5 U 
 
 
 O ~4 
 
 (1 
 
 « 
 
 01 x 
 
 0". 
 
 
 
 u w c 
 
 O 
 
 x; 
 
 
 
 
 
 •rl 
 
 
 
 
 ■H L. 0- C.-H 
 
 
 •rt c 
 
 c 
 
 y 
 
 ■H CJ 
 
 
 3 
 
 
 
 a 
 
 a 
 
 
 
 •H 
 
 
 4-1 
 
 
 «J 
 
 (J! 
 
 01 IT 
 
 JZ, 
 
 ^ 
 
 
 X M >,<4-i X CI 
 
 
 x; 
 
 o w 
 
 o 
 
 13 3 
 
 
 u e 
 
 
 
 y -r 
 
 C oi 
 
 i-. 
 
 
 
 ■y ~o t; 
 
 ■y £ 
 
 c 
 
 
 1/1 
 
 r 
 
 •"• O 
 
 4-1 
 
 *J 
 
 
 >- C 4-i c C 
 
 
 5 14 
 
 -H 
 
 c - 
 
 C 
 
 
 11 u 
 
 x 
 
 ■H -r-i 
 
 c in 
 
 c 
 
 
 
 -h -h a 
 
 ■H O 
 
 
 
 3 
 
 
 £ — 
 
 ■^1 
 
 
 
 1 13 l- -m -3 
 
 
 Cl 
 
 -. n 
 
 4J -H E 
 
 c 
 
 
 c 
 
 i> 
 
 r4 
 
 
 
 X 
 
 
 
 o --< o 
 
 o u 
 
 c 
 
 
 
 i: 
 
 ■n »4- 
 
 5 
 
 c 
 
 
 u ii p -4 c £ a 
 
 
 'C <J" 
 
 •3 
 
 TjH O 
 
 -J c • 
 
 
 C >H 
 
 
 •H 13 
 
 c 
 
 C 
 
 c 
 
 
 •h n c 
 
 ■H r3 
 
 •H 
 
 
 c 
 
 t,i 
 
 '/. -rH 
 
 
 •H 
 
 
 13 3 ~ -i s- o x; 
 
 
 r: 3 
 
 •-I C 
 
 1 4J 
 
 •n c 
 
 
 x 
 
 X 
 
 r4 > 
 
 01 
 
 
 it 
 
 
 >H > v. 
 
 »— l CJ 
 
 /: 
 
 
 y 
 
 r. 
 
 K O 
 
 V4 
 
 
 r- 
 
 ij jj x: -. c u k 
 
 
 IB 
 
 •-i H 
 
 4J q ii 
 
 >. 
 
 r. 
 
 rue 
 
 r.. 
 
 ■a 
 
 j. 
 
 CJ 
 
 r. 
 
 c a 
 
 c. « 
 
 ■y 
 
 
 
 •rH 
 
 C 
 
 3 
 
 Ci 
 
 u 
 
 x; c l, r -7 ,- --4 
 
 
 E <t 
 
 IB 
 
 cj C N 
 
 I-. 01 C 
 
 = 
 
 o I-. 
 
 u 
 
 x c 
 
 r. a 
 
 a x 
 
 c 
 
 X 13 
 
 ;< 
 
 ■H 
 
 
 o 
 
 )-. 
 
 ecu 
 
 u 
 
 
 
 C- r u,i< oh 
 
 
 a 
 
 > u> 
 
 cp m 
 
 It 13 
 
 "C 
 
 -1 CJ 
 
 x 
 
 a s 
 
 rs C 
 
 C. 
 
 
 
 C 
 
 t >■ 
 
 cj cr 
 
 3; 
 
 
 ^~ 
 
 4-> 
 
 LI 
 
 o- 
 
 c 
 
 x: 
 
 -j it tr. x; 
 
 
 o c 
 
 it y 
 
 C 
 
 <T »H » 
 
 3 
 
 X 3 
 
 c 
 
 y 
 
 
 
 
 
 ■H 
 
 LT^ 
 
 c 
 
 
 p. 
 
 ^. 
 
 tf. 
 
 4-1 
 
 c 
 
 (3 
 
 4J 
 
 1m ^ c U C 3 
 
 
 u y 
 
 t: 
 
 v. C o <*• 
 
 K C C 
 
 CI 
 
 - j cr c 
 
 c 
 
 !/) M 
 
 w 
 
 E3 
 
 y 
 
 c c 
 
 C -H 
 
 t: 
 
 c. 
 
 ■*- 
 
 
 •-4 X" 
 
 
 y 
 
 o 
 
 a -^ n c 2- 
 
 
 
 a -i 
 
 C. "4 x; • 
 
 O ■" 
 
 I-. 
 
 11 
 
 o 
 
 13 C7> C C 
 
 
 
 13 
 
 IB •" C 
 
 13 TJ 
 
 c 
 
 c 
 
 u: 
 
 ^" 
 
 
 
 y 
 
 K 
 
 
 H 13 CTi 
 
 
 13 o 
 
 u c 
 
 c: 4-i 4J O 
 
 CJ u.i -H 
 
 X 
 
 K C 
 
 
 3. 
 
 c 
 
 4J 
 
 iH 
 
 y 
 
 ■M C 
 
 ^3 
 
 
 
 c 
 
 v. 
 
 '^J 
 
 IC -H 
 
 i". 
 
 ■H 
 
 j:'. 
 
 3 O " C C • 
 
 
 
 
 13 c. 
 
 3 C 4J 
 
 O 4J 
 
 II 
 
 C -H 
 
 0. 
 
 ■- 
 
 ■1 1 CJ 
 
 
 
 IB 
 
 
 
 i3 — 
 
 CJ 
 
 •H 
 
 •f-1 
 
 c 
 
 c 
 
 i-i ;.'. 
 
 3 
 
 •^ 
 
 
 
 >CJ O ■" 'I • 
 
 
 c 
 
 
 4-1 tl CJ 
 
 C -> C 
 
 u 
 
 a 
 
 c 
 
 14 ~ 
 
 oi r*. 
 
 a 
 
 a 
 
 Of 
 
 14 3--R 
 
 ^ 
 
 K 
 
 c: 
 
 f:. 
 
 
 s 
 
 E 
 
 
 (3 
 
 — V, S-. = 
 
 
 1m 13 
 
 n c: 
 
 c c 
 
 U- »~4 
 
 c 
 
 01 
 
 
 G C 
 
 «■. < 
 
 u 
 
 K 
 
 a 
 
 1, O K 
 
 C 
 
 CJ 
 
 c^ 
 
 
 T 
 
 r>. 
 
 
 'C 
 
 x: y c o >44 ii 
 
 0- 
 
 c -y 
 
 »: 3 
 
 SZ O "i « 
 
 W •" 3 
 
 X 
 
 44 T 
 
 44 
 
 XI .J 
 
 a 
 
 y 
 
 3 
 
 H 
 
 £ c CJ 
 
 J. 51 
 
 
 
 
 
 4-> 
 
 •H 
 
 to >, 
 
 if. 
 
 c 
 
 
 410 0-1 -H 0' » 
 
 r 
 
 r. in 
 
 c c 
 
 uEcn 
 
 ■h E 
 
 4.' 
 
 ■H r-1 
 
 •H 
 
 41 
 
 i-i x 
 
 ti 
 
 
 
 
 4J ^ r- 
 
 y 
 
 k 
 
 k. 
 
 •H 
 
 4-1 
 
 E. '- 
 
 E 
 
 4-I 
 
 f 
 
 c. y - o 3 • 
 
 o 
 
 4-1 -i-i 
 
 •H -H 
 
 3 H C 
 
 C H 
 
 3 
 
 a 
 
 r. 
 
 U 
 
 o. y 
 
 c 
 
 
 
 4J 
 
 kCh 
 
 m 
 
 --, 
 
 Cj 
 
 
 r-1 
 
 V, 0! 
 
 K 
 
 13 
 
 
 
 r -i-i c in r,-. 
 
 r> 
 
 11 
 
 y i-i 
 
 COCK 
 
 tl c. c 
 
 < 
 
 l" —4 
 
 ■#4 
 
 tl C 
 
 X H 
 
 C x: 
 
 c 
 
 it X c 
 
 n c 
 
 X 
 
 X 
 
 -" 
 
 3 
 
 O 3 
 
 c 
 
 — ■ 
 
 v. 
 
 <- £. -j K r- it • 
 
 c. 
 
 i: tp 
 
 C <3 
 
 i- C X <t 
 
 (-■ I'. H4 
 
 : 
 
 M iij 
 
 X 
 
 M i-4 
 
 CJ ^ 
 
 G 
 
 -y 
 
 ■H 
 
 n u-» o 1 .r 
 
 H '< I 
 
 01 
 
 01 
 
 M 
 
 r. 
 
 y c 
 
 -J 
 
 
 14.; 
 
 O C " h i; > : 
 
 r-l 
 
 l-t rj 
 
 c > 
 
 o -3 x; 
 
 O IB 
 
 v. 
 
 
 h a 
 
 a 
 
 
 H i-l 
 
 v_1 
 
 
 *c ij 
 
 fc- 
 
 
 U >-i 
 
 M 
 
 
 1-4 L, 
 
 I. 
 
 r_; 
 
 t~ 
 
 
 2 
 
 t~ u; 
 
 L J 
 
 < 
 
 u r 
 
 Hi 
 
 
 « < 
 
 
 ^ 
 
 e. ?. 
 
 I-I 
 
 «c 
 
 Oj 
 
 ■J 
 
 W 
 
 ™ L. 
 
 >^ 
 
 c. 
 
 O ^/ 
 
 < 
 
 c 
 
 e: 7' 
 
 r 
 
 e 
 
 
 < 
 
 
 2 
 
 
 u 
 
 
 < 
 
 
 j 
 
 
 E-i 
 
 
 c 
 
 
 I/'. 
 
 
 c 
 
 
 1-4 
 
 
 c; 
 
 
 Q 
 
 E-i 
 
 
 c. 
 
 
 C C 
 
 fc-, 
 
 V. 
 
 c 
 
 C E 
 
 c 
 
 c 
 
 2 0. 
 
 
 
 M 
 
 < 2: 
 
 *I 3 
 
 c 
 
 C. 
 
 M 
 
 < < 
 
 '. : 
 
 c 
 
 H ~ 
 
 o; ti 
 
 M 
 
 2 
 
 <-. O 
 
 .j j 
 
 E< 
 
 5 tt 
 
 c c 
 
 tr. 
 
 c- 
 
 o c 
 
 o; 
 
 X 
 
 E. f-1 
 
 c a 
 
 i* 
 
 t4 
 
 £-, ci: 
 
 X o 
 
 
 -. Zi 
 
 Y. 
 
 K < 
 
 c 
 
 — C 
 
 H 
 
 '-^ ZT 
 
 H 
 
 < c 
 
 *^ 
 
 -- CJ 
 
 "-i 
 
 r_ 
 
 v: u. 
 
 
 
 
 
 ■J t 
 
 r m 
 
 < m 
 
 U ^4 
 
 M C, 
 
 CJ 1 
 
 c - 
 
 C rz 
 
 C : 
 
 ~ c 
 
 
 
 
 
 
 
 ___; 
 
 
 
 
 
 
 
 
 
 
 s 
 
 
 
 01 
 
 
 44 
 
 
 X 
 
 
 
 01 
 
 
 
 > 
 
 
 ti 
 
 
 
 
 -3 x: 
 
 
 
 
 
 
 U4 
 
 ■. 
 
 'J 
 
 
 
 y 
 
 
 G I, 3 
 
 
 'J 
 
 
 
 04 
 
 
 C 
 
 C - 
 
 
 c 
 
 
 '3 
 
 
 01 
 
 
 
 
 
 
 C 
 
 u = 
 
 : 
 
 
 
 •r4 
 
 
 01 u O X 
 
 
 o 
 
 
 
 tl 
 
 
 X 
 
 ■H 4J 
 
 
 Ci E 
 
 T3 
 
 o 
 
 
 C V-. 
 
 
 
 D04 
 
 
 
 
 * 
 
 
 
 
 
 
 S it 
 
 
 X 
 
 
 
 G X 
 
 t; 
 
 
 > 3 
 
 
 > C P' 
 
 a 
 
 r— 1 
 
 
 01 it 
 
 
 
 c c 
 
 * 
 
 
 y 
 
 C y 
 
 01 
 
 
 13 
 
 c 
 
 
 t4 E » 
 
 
 u 
 
 
 
 > X 
 
 3 Vi f3 
 
 0) 
 
 0) O 
 
 
 '3 C C 
 
 ^H 
 
 rt i— i 
 
 
 3 O 
 
 
 ~ 
 
 ■•-> 
 
 : 
 
 
 o 
 
 „ 3 
 
 G 
 
 
 
 ■n 
 
 
 g x: 14 o 
 
 
 
 
 '"J 
 
 ■H 31 
 
 C C k, 
 
 1-1 
 
 ~ c 
 
 
 x: c 
 
 = 
 
 C 
 
 
 01 
 
 
 14 
 
 .J ,-> 
 
 "41 
 
 
 01 
 
 G- 
 
 3 
 
 a 
 
 Ci 
 
 
 
 44 G G -H 
 
 
 X 
 
 
 G 
 
 3i-H 
 
 h fc. 0,1 <y 
 
 ' 
 
 X 
 
 
 X- -H 
 
 
 C 14 
 
 
 O 
 
 
 c 
 
 C t*. 
 
 
 
 
 
 13 u 
 
 ~t 
 
 X 
 
 c. 
 
 >-, 
 
 
 ■nlJU 
 
 
 
 
 
 r 4 
 
 CJ 
 
 ij r: ti 
 
 X 
 
 > flj 
 
 
 'd ti y 
 
 >- 
 
 ■r- J 
 
 
 ;J 1*4 
 
 
 3 
 
 C H 
 
 t 
 
 
 3 
 
 
 
 r 
 
 44 
 
 ■-4 
 
 G 
 
 
 >.X -H 
 
 
 ■r. 
 
 u 
 
 ^4 
 
 01 C 
 
 -: -r< 3 
 
 '.- 
 
 tl 
 
 
 "J -H 0> 
 
 i— ( 
 
 c 
 
 
 
 
 
 
 -< 
 
 
 
 
 rji 
 
 > 
 
 
 >■ >,X Z-. 
 
 
 P. 5 CC 
 
 
 X 
 
 
 
 c 
 
 CC4 
 
 -• ^ — 
 
 
 C i-i 
 
 
 c -n r 
 
 
 
 L C 
 
 
 
 
 >, 
 
 X 
 
 * 
 
 
 >. 
 
 C = 
 
 
 G 
 
 11 H 
 
 .. X 
 
 
 h GO 
 
 
 > 
 
 144 
 
 Ih 
 
 13 O 
 
 C 1 j 
 
 o 
 
 Vi c 
 
 
 13 r 3 
 
 4-1 
 
 CJ 'J 
 
 
 o y 
 
 
 c 
 
 O "^ 
 
 ; 
 
 
 V- 
 
 --4 
 
 G 
 
 > 
 
 -r, ^4 
 
 ■-1 
 
 
 C >■ > X 
 
 
 
 
 y 
 
 X --4 
 
 >. L. 4j x; 
 
 4J 
 
 o 
 
 
 E C Md 
 
 CJ 
 
 y 
 
 
 V, CJ 
 
 
 13 
 
 — G 
 
 t 
 
 
 •H 
 
 0" 01 
 
 X 
 
 PC 
 
 o rr 
 
 G O 
 
 
 1/1 '-4 -H 11 
 
 
 "3 
 
 E 
 
 C 
 
 14 4-1 
 
 S-T.C4J 
 
 ," 
 
 *-' •■ 
 
 
 L Cm 
 
 —1 
 
 ti >. 
 
 
 -4 01 
 
 
 
 *0J 
 
 r 
 
 
 G 
 
 3 f. 
 
 44 
 
 . X 
 
 G U 
 
 K 
 
 
 -i a c- 
 
 
 c 
 
 14 
 
 c 
 
 rj o 
 
 ■-. y o 
 
 4-1 
 
 tl 
 
 
 O 1 k. 
 
 c- 
 
 >4«H 
 
 
 -J 
 
 
 O 
 
 y 3 
 
 
 
 
 
 
 01 
 
 0- -rt 
 
 0- .-4 
 
 
 •t; --i oi 
 
 
 13 
 
 G 
 
 
 
 cx 
 
 c n 
 
 11 
 
 c 
 
 
 0' t. -H 
 
 u 
 
 01 O 
 
 
 *-' *- 
 
 
 01 
 
 X 1-4 
 
 u 
 
 
 C_ 
 
 > - 
 
 «— 1 
 
 -4 E 
 
 " 44 
 
 E G 
 
 
 3 13 01 
 
 
 E 
 
 X 
 
 
 T 
 
 Si t4 ■•* 
 
 
 l-i u 
 
 
 13 CJ 
 
 c 
 
 y 
 
 
 CO 
 
 
 3 
 
 G 01 
 
 ... 
 
 
 tl 
 
 X5 G 
 
 —4 
 
 G 
 
 C 
 
 14 
 
 
 EX 01 
 
 p.. 
 
 E 
 
 
 -3 
 
 'Cv X 
 
 c. c c 
 
 13 
 
 C H 
 
 
 E o x: 
 
 01 
 
 c CI 
 
 
 14 x; 
 
 
 
 y x 
 
 -r4 
 
 
 
 3 
 
 13 
 
 44 
 
 i4-i rj 
 
 G V4 
 
 
 CJ O O -4 
 
 y 
 
 C 
 
 4J 
 
 
 c a 
 
 V. 13 4J - 
 
 
 E C 
 
 
 ci r. y 
 
 
 x: h 
 
 
 01 
 
 
 V. 
 
 c 
 
 ^H 
 
 
 l-i 
 
 01 01 
 
 
 » L 
 
 c . 
 
 4J C 
 
 
 y U 4-i 
 
 ■#4 
 
 o 
 
 3 
 
 c 
 
 3 .-4 
 
 n. E 
 
 C 
 
 
 p. 
 
 u 
 
 V. 
 
 4J c 
 
 p.. 
 
 1-, -r. 
 
 
 5! 
 
 c 
 
 
 
 
 
 C 
 
 01 
 
 : >. 
 
 a 
 
 
 r- 
 
 tl O V4 
 
 
 
 C. 
 
 -^ 
 
 It 
 
 c -y t i-i 
 
 4J 
 
 O CJ 
 
 X". 
 
 t: V4 or 1 
 
 CJ 
 
 E 
 
 E 
 
 O -1 
 
 
 01 
 
 •»4 C 
 
 tl 
 
 
 tl 
 
 04 c 
 
 13 
 
 11 01 
 
 > o: 
 
 4- >- 
 
 c 
 
 :. > 4J c 
 
 
 
 IB 
 
 c 
 
 
 >, 
 
 4J it 
 
 
 to 
 
 o 
 
 s, o c 
 
 'J 
 
 >. 
 
 
 
 01 J3 
 
 
 3 
 
 13 
 
 -J 
 
 
 -j 
 
 CI c 
 
 x- 
 
 >4 
 
 44 
 
 C l-i 
 
 c 
 
 oi c 
 
 4-> ■ 
 
 
 --4 
 
 c 
 
 >4 O 
 
 - 4-1 
 
 --J 
 
 C w 
 
 I, 
 
 tl H 
 
 3 
 
 X. 
 
 y 
 
 3 it 
 
 
 
 ~j 
 
 1, 
 
 
 
 4-1 " 
 
 
 G C 
 
 •-! >^ 
 
 G it 
 
 •H 
 
 01 G O 
 
 
 O 
 
 
 G 
 
 a y 
 
 E w rj 
 
 01 
 
 3 
 
 p: 
 
 x- y 
 
 
 
 00 
 
 4J 
 
 
 G 
 
 CJ IP. 
 
 c 
 
 — 
 
 G 
 
 0^ 
 
 * 
 
 s X 
 
 •-< C 
 
 .-I ^4 
 
 y 
 
 C4IH 1( 
 
 ■3 
 
 t-i 
 
 c 
 
 ■H 
 
 rH 13 
 
 C « — 4 X" 
 
 c 
 
 
 c 
 
 f. iB 
 
 c 
 
 : 
 
 >, 
 
 O 01 
 
 P. 
 
 XC 
 
 x C. 
 
 5 
 
 p.. 
 
 XI 
 
 X 01 
 
 
 y 
 
 ■i~ 
 
 it 3 
 
 -H 
 
 X — i t 
 
 G 
 
 
 a 
 
 01 
 
 3 -H 
 
 
 j_l 
 
 01 -H 
 
 '. 
 
 x: -H- o 
 
 X 
 
 cr 
 
 o: 
 
 jr. ci 
 
 K 
 
 y 
 
 o. -. 
 
 
 £ 
 
 4-i 
 
 G 
 
 a 
 
 •■ 
 
 — . 01 
 
 > X 
 
 cc 
 
 4-1 > E 
 
 4-1 
 
 X 
 
 > 
 
 3 
 
 X "J 
 
 '. ■- E ;, 
 
 - 
 
 > 
 
 
 4-1 i-H *H 
 
 4J 
 
 n -i 
 
 
 y i 
 
 jr^ 
 
 
 _c 
 
 c. 
 
 CJ 
 
 
 - L, 
 
 i- 
 
 = o: 
 
 S .'. 
 
 ■r- 3 
 
 •-- 
 
 C -H 
 
 3 
 
 y 
 
 ■ri 
 
 -i 
 
 C G 
 
 O 4J rt 13 
 
 4> 
 
 
 o 
 
 07.-0 
 
 
 C 13 
 
 0. 
 
 CJ 
 
 v. 
 
 c 
 
 C- \~ 
 
 C 4. 
 
 £ 
 
 " f . 
 
 G 
 
 04 CJ 
 
 .', "3 
 
 3 01 
 
 14-, 
 
 C G 3 G 
 
 r-( 
 
 
 3i CJ 
 
 o £. 
 
 l-i O X 
 
 J 
 
 n; V. 
 
 x: 
 
 C C 
 
 01 
 
 H H 
 
 ST. 
 
 01 u 
 
 CJ 
 
 PC 
 
 N G 
 
 4-1 
 
 u 
 
 rr 
 
 G X 
 
 X 
 
 G C 
 
 " ^4 
 
 0' C 
 
 G 
 
 ig i4 c>4 
 
 G 
 
 -.1 
 
 
 . 
 
 C E 
 
 tt- (3 4) W 
 
 V- 
 
 E r., 
 
 4J 
 
 O U -H 
 
 H 
 
 -. ' 
 
 4J 
 
 H C 4J 
 
 u 
 
 - ... 
 
 
 = 
 
 CJ 
 
 01 G 
 
 >J 
 
 = G 
 
 O 'J 
 
 G > 
 
 ^ 
 
 0. *4-' 13 
 
 >4 
 
 HI 
 
 13 
 
 ■H 
 
 > H 
 
 CM 
 
 H 
 
 
 o 
 
 1-q 
 
 
 Eh 
 
 
 E-i 
 U op: 
 
 rt 
 
 E-i £-' E- 
 T. a. :i 
 
 M C i-l 
 
 c 
 r: w c 
 
 u X 
 
 flt.>4 
 
 W i-l ~ 
 P" E-i < 
 W X 
 1 3 E-i 
 
 k r < 
 
 
 >4 
 
 01 • 
 
 
 (S 
 
 D IB 
 
 
 w 
 
 
 
 s 
 
 
 
 a 
 
 X 
 
 
 
 
 
 PX 
 
 
 
 
 
 
 
 
 a 
 
 2 
 
 o 
 
 Q 
 
 
 
 
 
 
 ^j- 
 
 X 
 
 X 
 
 
 
 
 
 
 re 
 
 03 
 
 E- 
 
 
 
 
 
 
 < 
 
 c: 
 
 H 
 
 
 
 
 
 or. 
 
 u 
 
 w 
 
 44 
 
 
 
 Q 
 
 
 >^ 
 
 c 
 
 > 
 
 s 
 
 
 w 
 
 W 
 
 
 i- 1 
 
 > 
 
 2; 
 
 
 « 
 
 U 
 
 
 CJ 
 
 
 M 
 
 t-J 
 
 
 M 
 
 K 
 
 
 u 
 
 Q 
 
 
 X 
 
 
 t_3 
 
 C 
 
 
 c: 
 
 X 
 
 E-i 
 
 
 
 I c 
 
 4, 
 
 ;C 
 
 < 
 
 .. 
 
 X 
 
 a 
 
 0". 
 
 ^_. 7' 
 
 1 
 
 X 
 
 fcj 
 
 X 
 
 W 
 
 J2 
 
 \ 
 
 ' ■, '■ ■ 
 
 J 
 
 o 
 
 00 
 
 c 
 
 E-i 
 
 M 
 
 V 
 
 H '.' 
 
 X 
 
 c 
 
 
 
 
 71 
 
 ;' 
 
 ii§ 
 
 L, 
 
 X 
 
 : • 
 
 c ■ 
 
 X 
 
 r" 
 
 C 
 
 w 
 
 < 
 
 ;i 
 
 T' 
 
 J 
 
 (J 
 
 ;i 
 
 z cj 
 
 >4 
 
 M 
 
 ;■_ 
 
 c 
 
 IG 
 
 Eh 
 
 > 
 
 x n< 
 
 on 
 
 Q 
 
 C 
 
 <J 
 
 ft. 
 
 oo 
 
 CO 
 
 x' 
 
 G 
 
 
 u 
 
 it 
 
 X 
 
 o' 
 
 'O 
 
 
 1 
 7Z 
 
 
 C 
 
 
 CJ >4 
 
 
 K 
 
 
 Z < 
 
 
 H X 
 
 
 o 
 
 04 
 
 :. a 
 
 
 e? < 
 
 CJ 
 
 0J u 
 
 fc. fcj 
 
 E« o 
 
 C Eh 
 
 
 
 O 
 
 >4 a 
 
 Z Q 
 
 < X 
 
 X X 
 
 X E-4 
 
 X X 
 
 Lr5 
 
 CJ X 
 
 x c 
 
 M X 
 
 x « 
 
 G 01 
 
 CJ E> 
 
18 
 
 CERL HDB Recommendations 
 
 C\J 
 
 Etf HMKSOOaw^BK ?ti&acs 
 
 itj'^3!»5a^ir2«ai - 
 
 IC • 
 
 
 
 
 .-t 
 
 
 C 
 
 
 
 
 
 
 
 x: - o 
 
 CJ 
 
 
 
 
 u o 
 
 n 
 
 
 u-* 
 
 
 M rH S-l 
 
 
 CD 
 
 C • -1 
 
 
 C •" 
 
 
 
 x: 
 
 <3 c 
 
 
 u.. i_ 
 
 
 x> 
 
 3 
 
 
 W C 
 
 -H 
 
 
 ■J-. 
 
 
 a 
 
 ■*-< 
 
 c 
 
 •H '3 
 
 
 a zr. a 
 
 
 
 U 
 
 X 
 
 > 4) 3 
 
 o 
 
 
 - 
 
 r,. 
 
 c 
 
 jj C T 
 
 c 
 
 c 
 
 C 3 •• 
 
 
 
 tj 
 
 CJ TJ TJ C 
 
 o 
 
 o 
 
 E -h a c 
 
 
 >; 
 
 a > *J u 
 
 >- 
 
 
 #H -X n; "-' 
 
 Xi 
 
 .n 
 
 (XT3 
 
 
 u 
 
 E C C • 
 
 
 r) 
 
 ^ e 
 
 :.. x 
 c 
 
 S 3 
 
 C 
 
 r l 
 
 '- 
 
 3 
 
 ^. 
 
 -J 
 
 c a 
 
 •- 
 
 
 r; 
 
 4J 
 
 
 
 c 
 
 ill 
 
 y 
 
 -»j 
 
 Is 
 
 "' 
 
 C 
 
 c 
 
 «lJ 
 
 "J 
 
 ,-, 
 
 * 
 
 f, 
 
 ~ 
 
 rt 
 
 £. 
 
 c - 
 
 r> 
 
 rc x" c ■>- 
 
 c 
 
 X C ■ V- E 
 
 •-- 
 
 17 C 
 
 ..) 
 
 c - ■ — - 
 
 
 V, r-( *J • 
 
 
 
 
 
 
 ~ E O • 
 
 n 
 
 C 'r- U O 
 
 > 
 
 o v. >■ 
 
 
 > t; xj 
 
 
 
 
 X C -H = 
 
 >' 
 
 u x r-i a 
 
 V, 
 
 o x- 1 --i n 
 
 o • 
 
 ^1 .C -H 
 
 3 c* 
 
 c c c .V 
 
 'S X • 
 
 u: c c ;: 
 
 ■r- CO 
 
 L' i- K 
 
 .« x i 
 
 C <J '-I 
 
 3 
 
 X. - ' = 
 
 C V ■ J 
 
 -; T 
 
 ■P c 
 
 x h 
 
 !> 
 
 ■j :: C C 
 
 <•: X 
 
 4. 1 
 
 r. x. ^h 
 
 jj' ' tr 
 
 " 'C '-'. £ 
 
 a £ o 
 
 "_ J; M -H 
 
 > 3 -H 
 
 •n C •"- 
 
 a c h-i 
 
 :'. X "- x 
 
 f~* 
 
 .- O O C 
 
 a .<« c 
 
 O n rx V. 
 
 u c c 
 
 - .'-. (.1 'XI 
 
 I; C <.'. -H 
 
 '.- ■-. "J r-J 
 
 c ~ c c 
 
 
 c <X 
 
 
 C -r* <-* 
 
 
 •C. M-) ft >i 
 
 
 
 
 
 
 XI -H 
 
 
 C C E w- 
 
 
 £ f Vi 4J 
 
 
 jj -j c r. 
 
 
 •r- +J ', 
 
 
 E 
 
 
 xi v. .t c 
 
 
 C rH 
 
 
 K .C K^ O 
 
 4 
 
 U u> 
 
 'J) 
 
 fl 
 
 i-l 
 
 u-, a x< 
 
 c 
 
 U V) «s 
 
 . 
 
 K C 3 ""J 
 
 " 
 
 H fi -■ CI U 
 
 
 •J 
 
 
 
 
 
 < 
 
 
 
 
 
 M 
 
 
 
 
 
 e^ 
 
 
 
 E- 
 
 
 x 
 
 
 
 — ' 
 
 
 w 
 
 
 
 f" 
 
 
 D 
 
 
 
 
 
 O 
 
 c 
 
 
 y " 
 
 
 < cr. 
 
 
 r 
 
 C C 
 
 
 H* 
 
 < 
 
 > 
 
 3 < 
 
 fX 
 u. Q 
 
 a o 
 
 C C 
 
 H C 
 
 p: 
 
 E- 
 
 'X 
 
 r K: U 
 C 14 >i 
 
 rx 
 
 
 S G 
 
 a -z -a 
 z o □ 
 
 v. 
 
 < 
 
 Z I- 
 
 c c 
 
 < t; z 
 
 tf W t-t 
 
 E- 
 
 ~ 
 
 E-; & 
 
 
 K 
 
 ^7. 
 
 
 
 • » • 
 
 C 
 
 < 
 
 
 " 
 
 M C- 
 
 tr. 
 
 ' 
 
 
 
 
 z 
 
 
 ShS 
 
 §2 
 
 H 
 
 
 
 
 c 
 
 HI 
 
 
 M HI O 
 
 C 
 
 r 3 
 
 
 
 
 fX 
 
 
 M c u. 
 
 |H 
 
 Hi 
 
 m 
 
 
 U 
 
 u 
 
 CJ 
 
 
 a o 
 >: z > 
 
 (X! M < 
 
 o 
 ,J 
 
 CJ 
 IS1 
 
 < 
 e. 
 
 cj 
 
 
 U 
 
 C £- 
 
 Z 'X 
 
 hi n 
 
 E-" » 
 
 6 
 
 cx, 
 
 ^ 
 
 I A FI 
 OFF-I 
 DISPI 
 
 < 
 
 cj 
 
 o 
 
 5 
 o 
 
 z t- 
 
 C * 
 cj < 
 
 X 
 
 rx 
 
 HI 
 
 c. 
 
 z 
 
 Hi 
 
 K 
 
 Q 
 
 
 d. 
 
 
 
 , . 
 
 3 
 
 a 
 
 >• 13 X) 
 
 c 
 
 rc 
 
 x! 
 
 U TJ 
 
 CJ 
 
 n 
 
 
 
 
 
 
 
 co 
 
 » 
 
 w 
 cr. 
 < 
 
 < C XI 
 E- O (J 
 
 < y h 
 
 D CS IX 
 
CERL HDB Recommendations 
 
 19 
 
 immediately to elimination of unsuitable choices based on the 
 technical features. 
 
 The NBS report suaoests drawing distinctions in three broad 
 classes of system applications: formatted data processing/ 
 structured text searching/ and personal text handling. The needs 
 of the HDB fall into the class of structured text orocessinq. in 
 the following excerpt from the NBS report/ the items in 
 parentheses refer to the characteristic features listed in the 
 Quest ionnai re . 
 
 repr 
 sear 
 
 wher 
 
 segm 
 
 Exam 
 
 abst 
 
 sect 
 
 ther 
 
 d e s c 
 
 the 
 
 inco 
 
 desc 
 
 reco 
 
 voca 
 
 chec 
 
 Or e 
 
 may 
 
 Beca 
 
 unac 
 
 Eng I 
 
 Beca 
 
 reco 
 
 the 
 
 of f- 
 
 esse 
 
 a co 
 
 se le 
 
 judg 
 
 Str 
 esen 
 chin 
 e t 
 ent s 
 pies 
 ract 
 i ons 
 e mu 
 ript 
 text 
 nsi s 
 ript 
 rd 
 
 bu la 
 k h 
 Ise/ 
 
 be 
 use 
 cust 
 i sh- 
 use 
 rds 
 
 usu 
 line 
 nt i a 
 unt 
 cted 
 e th 
 
 uc tyred 
 t at i ve 
 g/ lega 
 he f i I 
 of t 
 of te 
 s/ pat 
 . To i 
 st be a 
 ion/ si 
 
 i n a s 
 tent w 
 ion may 
 by a 
 ry (E.2 
 is desi 
 
 any si 
 
 provi d 
 these s 
 omed 
 like ph 
 a se 
 that wo 
 al 10 
 
 pr i nt i 
 
 t . Ftor 
 
 (F.3.C) 
 
 by a 
 
 e des i r 
 
 t ex 
 o 
 I te 
 e r 
 ex t / 
 xt s 
 ent 
 dent 
 
 tec 
 nee 
 egme 
 ith 
 
 be 
 set 
 .a) 
 red 
 gni f 
 ed 
 
 yst e 
 to 
 ras i 
 arch 
 uld 
 
 or 
 ng ( 
 eove 
 of 
 
 pro 
 abi I 
 
 t s 
 f 
 
 x t s 
 
 ecor 
 
 (1 
 
 egme 
 
 c la 
 
 ify 
 
 hn ig 
 requ 
 nt w 
 the 
 orov 
 
 of 
 wh i c 
 term 
 i can 
 as a 
 ms a 
 
 pro 
 
 ng i 
 
 ma 
 
 be e 
 
 30 
 F.2. 
 r/ t 
 th 
 pose 
 ity 
 
 ear c 
 bibl 
 earc 
 ds 
 000 
 nt s 
 i ms/ 
 ate 
 ue f 
 i rin 
 ould 
 i nte 
 ided 
 
 key 
 h th 
 
 for 
 t wo 
 
 val 
 re s 
 g ram 
 s de 
 y s 
 xcee 
 char 
 a) a 
 hese 
 e r 
 d se 
 of c 
 
 hi ng 
 i oq ra 
 hi ng / 
 consi 
 
 char 
 wou Id 
 par 
 xt re 
 or ab 
 q an 
 
 be 
 nded 
 
 by i 
 words 
 e use 
 
 acce 
 rd o 
 id se 
 pec i a 
 mi ng 
 emed 
 elect 
 di ng I 
 act e r 
 t hig 
 
 sy st 
 eco rd 
 arch 
 ont i n 
 
 is con 
 ph i c i 
 
 and s i m 
 st of 
 acters 
 
 be repo 
 agraphs/ 
 cord for 
 bre v i at e 
 exact ma 
 inconce i 
 f unct i on 
 ndex i ng 
 
 f rom a 
 r can i 
 pt abi I i t 
 ecu r r i ng 
 arch t er 
 I ly ai me 
 encode 
 essent i a 
 vo I umi 
 y I ong t 
 /second 
 h speed 
 ems sbou 
 s that 
 so that 
 ui nq the 
 
 cei v 
 
 nf or 
 i I ar 
 pr es 
 or 
 
 rt t 
 s 
 se I 
 d c 
 tch 
 vab I 
 . C 
 each 
 cont 
 nspe 
 y <e 
 
 i n 
 m (E 
 d at 
 d 
 
 I as 
 nous 
 o pr 
 term 
 i s 
 Id d 
 
 wou 
 a us 
 
 sea 
 
 ed ?»s 
 mat ion 
 
 uses 
 c r i bed 
 more > . 
 i 1 1 es / 
 tat ute 
 ect i on 
 ont ent 
 to al I 
 e and 
 ont ent 
 
 file 
 rol led 
 ct to 
 .^.f) . 
 
 text 
 
 .2.b). 
 
 users 
 
 forms/ 
 
 we 1 1 . 
 
 text 
 i nt on 
 i na Is / 
 
 also 
 resent 
 Id be 
 er can 
 r ch . 
 
20 
 
 CCRL HOB Recommendations 
 
 The data files of structured text searching 
 systems would be expected to be unchanging in 
 content and very large in volume. It would be 
 expensive to reorder or restructure them as new 
 data is received/ so it would be desirable for the 
 system to accept new data in any order (D.3). 
 Other desirable features would extend content 
 searching capability* for example by giving a 
 synonym facility ( E . ? . d ) or a presentation of 
 other terms that are conceptually related (E.2.e). 
 As in formatted data processing* tutorial aid is 
 desirable. In contrast to that application 
 however* full Boolean capability* optional report 
 formattinu* and optional ordering ar^ sugaested 
 here as desirable rather than essential. Only a 
 Boolean AND* allowing the conjunction of distinct 
 search terms* is imperative for user convenience* 
 to avoid a tedious selection from record subsets 
 found by individual terms. Optional formatting 
 and ordering may not be used often for such simple 
 structured output records as bibliographic 
 citations. A standard output presentation then is 
 generally sufficient* unless text fields become 
 numerous and frequently of marginal importance* 
 requiring more selectivitty to be niven the user. 
 
 The chart from the hbS report for cateqorizinq systems is 
 
 reproduced here as Table 3. Figure 1 shows just the entries 
 
 which have an x in the feature row cor r espond i ng to structured 
 
 text processinq systems. This figure shows in a compact format 
 
 the choices which on the face of it would be suitable for the HDB 
 
 application. Those systems marked with a "+ M are listed 
 
 specifically as allowing customer data bases to be added to a 
 
 "service" system. Also* two systems are included on this table 
 
 which are not mentioned in the \RS report. These are the CELDS 
 
 system and the EUREKA system currently in some stage of 
 
 development at the Urbana campus of the University of Illinois. 
 
 The SMART system is also indicated on this chart* though it does 
 
CERL HDB Recommendations 
 
 21 
 
 
 VIVOINTl 
 
 NOOLL 
 
 0003 KUSAS 
 
 sjuvis 
 n sands 
 
 HVTOS 
 X0S30HS 
 
 sfiia 
 
 N0O3H 
 SDWH 
 
 axvoan aon& 
 
 III XISHO 
 
 SSATIO 
 
 S3WLL*A - N 
 
 SISVN 
 
 3STW 
 
 90VHIW 
 
 VIVOIKIW 
 
 IXUOMDIW 
 
 TCMINCO H3XSVW 
 
 IA SHVW 
 
 in savw 
 
 AI XHVH 
 1USNI 
 
 aainbra: 
 
 8/SMI 
 
 09E/SWI 
 (d30)SW 
 
 savwi 
 
 ASdIS 
 SIHECTU 
 
 iavsiwi 
 
 e/sa 
 
 an 
 
 savm 
 scnvia 
 
 (Xauuso/eiHa) 
 
 ioomio 
 
 SWCD 
 
 sisva 
 
 1 I 
 
 C « ' 
 
 
 X 
 
 
 
 
 
 
 
 
 X 
 
 
 
 
 X 
 
 
 X 
 
 X 
 
 
 
 
 
 X 
 
 
 
 
 
 
 X 
 
 
 
 
 
 
 X 
 
 
 
 
 
 
 X 
 
 
 
 
 X 
 
 
 
 
 
 
 
 
 X 
 
 
 
 
 
 
 X 
 
 
 
 
 
 
 X 
 
 
 
 
 X 
 
 
 
 
 
 
 
 X 
 
 
 
 
 
 
 
 X 
 
 
 
 
 
 X 
 
 
 
 
 
 
 
 X 
 
 
 
 
 
 
 X 
 
 
 
 
 
 
 
 
 X 
 
 
 
 
 
 X 
 
 
 
 X 
 
 
 
 
 
 
 
 
 X 
 
 
 
 
 X 
 
 
 
 
 X 
 
 (O 
 
 a; 
 
 •H 
 
 X 
 
 
 
 
 
 </5 
 
 
 
 
 
 X 
 
 1 
 
 
 
 X 
 
 
 
 a 
 
 X 
 
 
 
 
 
 1 
 
 
 
 
 
 X 
 
 . 
 
 
 
 X 
 
 
 
 
 
 
 
 X 
 
 
 
 
 
 X 
 
 
 
 
 
 
 
 X 
 
 
 
 
 X 
 
 
 
 
 
 
 
 X 
 
 
 
 
 
 
 X 
 
 X 
 
 
 
 
 
 
 
 X 
 
 
 X 
 
 
 
 
 
 
 
 
 
 X 
 
 
 
 
 
 X 
 
 
 
 
 
 
 X 
 
 
 X 
 
 
 
 
 
 X 
 
 
 
 
 X 
 
 
 
 
 
 
 
 X 
 
 
 
 
 
 
 X 
 
 
 
 
 
 
 X 
 
 
 
 
 
 
 X 
 
 
 
 
 X 
 
 
 
 X 
 
 
 
 o 
 o 
 o 
 
 o 
 
 g 
 
 0- 
 
 o 
 r- 
 
 <v> 
 
 o 
 
 en 
 
 5 
 
 o 
 
 o 
 
 5 
 
 & 
 
 
 X 
 
 
 X 
 
 
 X 
 
 
 X 
 
 X 
 
 X 
 
 X 
 
 X 
 
 X 
 
 X 
 
 X 
 
 X 
 
 
 * 
 
 
 X 
 
 
 X 
 
 
 X 
 
 X 
 
 X 
 
 X 
 
 X 
 
 X 
 
 
 X 
 
 
 
 X 
 
 
 X 
 
 
 X 
 
 X 
 
 X 
 
 X 
 
 X 
 
 X 
 
 X 
 
 
 X 
 
 
 X 
 
 
 X 
 
 X 
 
 X 
 
 
 X 
 
 
 X 
 
 
 X 
 
 
 X 
 
 X 
 
 
 X 
 
 
 X 
 
 X 
 
 
 X 
 
 X 
 
 
 
 X 
 
 
 X 
 
 
 X 
 
 X 
 
 X 
 
 X 
 
 
 X 
 
 
 X 
 
 X 
 
 X 
 
 X 
 
 X 
 
 
 X 
 
 X 
 
 X 
 
 X 
 
 8 
 
 I 
 
 £ 
 
 
 X 
 
 
 X 
 
 
 X 
 
 
 X 
 
 X 
 
 
 X 
 
 
 X 
 
 
 X 
 
 
 X 
 
 
 X 
 
 
 
 X 
 
 
 X 
 
 X 
 
 
 
 X 
 
 
 X 
 
 
 X 
 
 X 
 
 
 
 X 
 
 X 
 
 
 
 X 
 
 X 
 
 
 X 
 
 
 
 X 
 
 
 X 
 
 
 X 
 
 X 
 
 
 
 X 
 
 
 X 
 
 
 X 
 
 
 X 
 
 X 
 
 
 
 X 
 
 X 
 
 
 
 X 
 
 
 X 
 
 
 X 
 
 X 
 
 
 
 X 
 
 
 X 
 
 
 X 
 
 
 X 
 
 
 X 
 
 
 X 
 
 
 X 
 
 
 X 
 
 
 X 
 
 o 
 
 -1-1 
 
 1 
 
 I 
 
 >H 
 
 CD 
 
 CO 
 
 Eh 
 
 S 
 O 
 
 CO 
 
 >H 
 
 CO 
 
 o 
 
 ie; 
 o 
 
 H 
 
 Eh 
 
 <: 
 
 Csl 
 H 
 
 O 
 
 o 
 
 i-5 
 
22 
 
 CERL HDB Recommendations 
 
 0) 
 
 o 
 
 P- 
 
 o 
 
 + 
 
 oo 
 
 b 
 
 o 
 
 VXD 
 
 + £ 
 
 CO 
 
 >-< o 
 
 
 CO « 
 
 g 
 
 PL, O 
 
 (XI 
 
 sa 
 
 H 
 
 H 
 
 o 
 
 W X 
 
 VD 
 
 Eh O CO 
 
 OO 
 
 >h O CO S pq PC W 
 CO K M O W <J K 
 
 g 
 
 P-, O CO O O i-h H 
 >h M < W X O Ph 
 
 w 
 
 M 
 
 O g S K CO CO CO 
 
 O CO 
 < Q 
 
 > ^ 
 M Ph 
 
 g 
 
 
 
 
 
 
 
 
 
 m 
 
 
 
 
 O 
 
 
 
 
 ■V 
 
 o 
 
 H 
 
 
 
 
 H 
 
 0) 
 
 
 
 
 
 
 CJ 
 
 
 
 > 
 
 
 
 
 M 
 
 o 
 
 
 
 u 
 
 
 
 
 n 
 
 a 
 
 
 
 V 
 
 
 O 
 
 
 
 
 
 
 
 o 
 o 
 
 4- 
 
 
 
 c_> 
 
 4- 
 
 
 — . vo 
 
 CO 
 
 
 
 < 
 
 co 
 
 hSJ 
 
 
 
 H 
 
 
 
 > 
 
 H 
 
 r— 
 
 Eh 
 
 on 
 
 S 
 
 o 
 
 e + 
 
 vo 
 
 iJ "^ o 
 
 on 
 
 o < o 
 
 g 
 
 O EH J 
 PC <C -3 
 
 W 
 
 M Q M 
 
 H 
 
 O— O 
 
 CO 
 
 H M 
 EH M 
 
 >H H 
 
 S O 
 
 CO 
 
 S 
 
 w 
 
 EH 
 CO 
 >h 
 CO 
 
 o 
 S2i 
 
 H 
 
 o 
 
 CO 
 
 w 
 
 Eh 
 O 
 
 Eh 
 O 
 
 D 
 
 § 
 
 CO 
 
 H 
 
 o 
 o 
 
 pq 
 
 PQ 
 
 
 
 
 
 
 
 
 
 (U 
 
 
 
 
 
 
 
 
 a 
 
 
 
 
 
 
 
 
 /.-; 
 
 
 
 
 •• 
 
 
 
 
 o 
 
 
 
 
 o 
 
 
 
 
 Ph 
 
 
 
 
 H 
 
 0) 
 
 
 
 
 
 
 
 
 
 
 
 
 
 o 
 
 c 
 
 
 
 
 
 
 
 pq 
 
 o 
 
 
 
 
 
 o 
 
 
 n 
 
 c 
 
 
 
 
 
 
 
 o 
 
 
 
 
 • • 
 
 
 
 o 
 
 
 
 
 u 
 
 
 
 
 L ^ 
 
 CO 
 H 
 
 
 
 •g 
 
 CO 
 
 hH 
 
 o 
 
 
 
 
 
 
 
 c— 
 
 Fn 
 
 
 
 
 
 
 00 
 
 W 
 
 
 
 
 HH 
 
 
 o 
 
 CJ 
 
 
 W 
 
 > 
 
 HH 
 
 
 vn 
 
 "-\ 
 
 C"J 
 
 P5 
 
 H 
 
 CO 
 
 
 00 
 
 <d 
 
 < ) 
 
 of 
 
 
 eh en 
 
 a 
 
 g 
 
 bH 
 < 
 
 <^ 
 
 « 
 
 PQ <! 
 
 o 
 o 
 
 P4 
 
 « 
 
 H 
 
 s 
 
 <E 
 
 pr. Eh 
 
 M 
 
 H 
 
 
 
 W 
 
 M 
 
 5; 
 
 O CO 
 
 H 
 
 o 
 
 o 
 
 CO 
 
 < 
 pq 
 
 b ^ 
 
 pq 
 
CERL HDB Recommendations 
 
 23 
 
 not meet the criteria required for inclusion in the NBS survey. 
 Tables A and 5 characterize the SMART and CELDS system features/ 
 while Table 6 characterizes the "ideal" system for an HDB. 
 
 6. The Choice of an Appropriate HDb System 
 
 The process of finding an appropriate information retrieval 
 system to be used for the HDF reduces first to the decision of 
 whether or not natural language queries are necessary and then* 
 if they are necessary* to deciding whether SMART is satisfactory. 
 If it comes close to beina satisfactory* then one can consider 
 how it can be implemented so that it is available to CFRL. 
 
 6.1 The Choice of Natural Language Query 
 
 There is a seemingly unanswered question of whether or not 
 natural language inquiry and fully automated language analysis 
 procedures are effective in a document retrieval environment such 
 as the HDB. Note here that we are assumina that the HUB task is 
 equivalent to document retrieval in the sense that the statement 
 content is similar to abstracts. However/ in the techniques now 
 being used for the HDB the title* author* and other citation type 
 of information is not used. The only thing used is the text of 
 the statement itself* the index is only used in the ANDOR 
 approach . 
 
24 
 
 CEPL HDB Recommendations 
 
 a 
 
 v-» 
 
 o 
 
 U> 
 
 >- 
 
 
 
 "O 
 
 
 « 
 
 t- 
 
 
 :.- 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 >-l 
 
 1= 
 
 
 
 
 
 
 I 
 
 z 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 Z 
 
 o 
 
 
 
 
 
 
 o 
 
 o 
 
 
 a 
 
 
 8« 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 HI 
 
 
 
 
 
 
 
 ;j 
 
 
 
 
 z§ 
 
 
 
 
 
 
 
 
 
 fa. 
 
 
 
 
 
 
 C-, 
 
 CO 
 
 
 
 
 
 2 
 
 r_, 
 
 
 
 
 
 
 
 
 
 
 
 
 O 
 
 
 
 
 
 
 b 
 
 Z 
 
 
 
 
 < 
 
 c- 
 
 
 < 
 
 
 Hi J 
 
 
 
 
 
 
 
 
 < 
 
 
 
 
 
 
 
 g 
 
 o 
 
 
 
 
 o 
 
 w 
 
 H 
 
 
 
 
 3 
 
 
 
 E- 
 
 
 
 
 
 ££ 
 
 o 
 
 
 
 
 o 
 
 
 HI 
 
 E-i 
 
 PH 
 
 
 o 
 z 
 
 c 
 
 > 
 
 jg 
 
 5 
 
 
 i 
 
 
 S3 
 
 
 
 
 B 
 
 
 
 
 o o 
 
 HI 
 
 
 
 
 z 
 
 
 g 
 
 c 
 
 Z 
 
 HI 
 
 
 
 Pm 
 
 
 
 
 
 w 
 
 E- 
 
 
 
 
 w 
 
 c: 
 
 
 
 
 
 
 o 
 
 Hi 
 
 H. EC 
 
 
 c 
 
 
 c_ 
 
 
 5. 
 
 
 E-i O 
 
 
 ^ 
 
 C3 
 
 ►J 
 
 
 
 
 C. - : 
 
 w 
 
 
 
 1 
 
 X 
 
 
 H 
 
 
 o 
 
 1 O £ Cd 
 = Z 1 5 
 
 <_> 
 
 fc: 
 
 
 
 
 C 
 
 
 > 
 
 
 § 
 
 Z 
 
 W 
 
 Pi 
 
 
 a z 
 
 W (_> 
 
 (3 
 
 < 
 
 .t. 
 
 
 
 
 w 
 
 u. 
 
 § 
 
 
 p< 
 
 o 
 
 '.1 
 
 
 
 o 
 
 Z 
 
 K 
 
 Hi 
 
 O 
 
 
 o o u 
 
 **r 
 
 EC 
 
 H 
 
 O 
 
 Q 
 
 c 
 
 
 § 
 
 
 
 j 
 
 w 
 
 
 
 >- 
 
 
 Z Q 
 
 O 
 
 S 
 
 iJ 
 
 >, 
 
 
 
 h. &: m z 
 
 ^j E 
 
 o 
 
 < 
 
 
 
 z 
 
 
 r- 
 
 >-l 
 
 10 Hi E U 
 
 w 
 
 o 
 
 E-i 
 
 Hi 
 
 
 < 
 
 
 HI K 
 
 H 
 
 e 
 
 
 
 < 
 
 Z 
 
 ht-f-r- 
 
 <i O 
 
 
 C 
 
 o 
 
 < 
 
 Hi 
 
 
 P 
 
 HI 
 
 
 Hif. KO 
 
 CO 
 
 £ 
 
 
 g 
 
 7'. 
 
 
 w 
 
 id iJ 
 
 En 
 
 K 
 
 Q 
 
 E 
 
 c 
 
 < < M E- 
 
 K H 
 
 3 
 
 
 HI 
 
 1 
 
 
 
 t/5 
 
 E~ 
 
 < 
 
 -J < E-> J 
 
 OK U< 
 
 
 h1 
 
 g 
 
 c- 
 
 I 
 
 O w 
 
 HI 
 
 s 
 
 
 
 | 
 
 
 PJ C J hi 
 
 o :- 
 
 a 
 
 
 U- 
 
 El 
 
 
 
 
 
 E* 
 
 z 
 
 J 
 
 s 
 
 
 
 U3 o 
 
 Z 
 
 O 
 
 § 
 
 E-i 
 
 < 
 
 w 
 
 k a. w c 
 
 U 5 O W 
 
 SS 
 
 o 
 
 K 
 
 g 
 
 Z 
 
 5 
 
 
 oo 
 i 
 
 E-i 
 
 < 
 
 R. — >" tt 
 
 W (X, to O 
 
 5 
 
 o 
 o 
 
 E 
 
 E-i 
 
 03 
 
 
 H 
 
 c 
 
 g 
 
 O Eh 
 
 S 
 
 (-H 
 
 « 
 
 M 
 
 5 
 
 
 0* t-i 
 
 
 5 
 
 p 
 
 
 8 
 
 
 cc 
 
 
 
 
 
 
 
 
 
 
 
 
 H 
 
 § 
 
 fe 
 
 S 
 
 3 
 
 • . . . 
 
 S3 
 
 E-- 
 
 £h 
 
 £> 
 
 o 
 
 
 
 . 
 
 
 . 
 
 z 
 o 
 
 
 
 
 
 
 
 . 
 
 H 
 
 h1 
 
 E 
 
 ?» 
 
 2 
 
 w 
 
 ■J 
 
 M 
 
 < 
 
 (JOT) 
 
 A. > 
 
 O 
 
 W 
 
 < 
 
 5t 
 
 >-• 
 
 3 
 
 0) 
 
 
 ,o o 
 
 o 
 
 * 
 
 p 
 
 o 
 
 "3 
 
 t. 
 
 
 V- 
 
 rH 
 
 C\J 
 
 00 
 
 ri 
 
 
 Cvj 
 
 on 
 
 
 -s 
 
 
 
 M 
 
 
 
 
 oj 
 
 
 
 
 
 
 
 
 t- ^ a 
 
 •^ ffi o 
 O 3 
 \C E Q. 
 
 C CO 
 
 Eh 
 
 Of 
 
 o fc' 
 
 M G 
 
 3 
 
 ^ o 
 
 *^ H 
 
 - i 
 
 z o <c 
 
 E- h a 
 
 ti ;-. £-• 
 
 x <-" fc3 
 
 HI H Z 
 
 k U M 
 
 C Hi E-i 
 
 •; hi c: 
 
 In 
 
CERL HDB Recommendations 
 
 25 
 
 8 3 
 
 < 
 
 O 
 
 ►3 
 
 tr> 
 
 o 
 
 w 
 
 < 
 
 w 
 
 M 
 
 O 
 
 £C 
 
 E»- 
 
 6-1 
 
 ■ 
 
 5g S 
 
 0) 01 ct tu 
 
 « o t « o o o 
 
 x ti 
 
 no w < 
 
 O Q u E-i 3= O 
 
 
 
 
 
 .-a .- O to CO t: 
 
 
 o 
 
 
 t~ ^ C »-i M HH 
 
 to 
 
 
 
 J hi 
 
 m »-i w a iJ x 
 
 t«4 wu. Q E-i to o o 
 
 £ 
 
 IP 
 
 >- 
 
 = K O lv r. Q ;- K 
 
 w r-i 5 o E o o si o w < 
 
 Sh t-, «J i.j (•: 
 <! S S S ~, C Ol w Jin 
 
 JKW^I-iWtlCKB 
 
 < 
 
 gr? 
 
 £ 
 H 
 
 £ 
 
 
 s 
 
 
 
 a. >; ►-. 
 
 E-i 
 
 3 
 
 r : Jb -5 
 
 CO 
 
 p c u; 
 
 Ol^OOOlJfcKitt-K 
 
 2Scqqk£;xw<-3 
 
 FlU 
 
 
 5^9 
 
 o; 
 
 ;c 
 C 
 
 S ID H 
 
 sp om in. 
 
 J £ 
 
 e 2 
 
 H<3 
 
 Eli 
 
 ft 3 
 
 ll. t»- t"> 
 o o 
 
 i-i «£ <x 
 
 [<- t: M 
 k >-■ a. 
 O « to 
 
26 
 
 CERL HDB Recommendations 
 
 §1 
 
 
 su 
 
 V o o o o 
 ;-. ?! z z a 
 
 Eh 
 
 u. 
 
 W 
 
 o 
 
 z 
 
 H 
 
 w 
 
 
 w 
 
 w 
 
 set 
 
 03 
 
 3 
 
 cq 
 H 
 
 1 
 
 3 
 
 
 w 
 
 
 
 
 w 
 
 J 
 
 
 
 
 •-3 
 
 eh e- 
 
 s s 
 
 
 
 
 
 
 z 
 
 
 
 
 
 5 
 
 ^ 
 
 
 
 
 
 1 
 
 Z 
 
 o 
 
 
 
 
 
 
 o 
 
 
 
 
 
 a 
 
 a 
 
 
 
 
 CJ >> 
 
 
 
 
 
 
 
 
 
 
 
 
 w 
 
 
 
 
 
 
 <: 
 
 o 
 
 
 
 2 
 
 CO 
 
 z 
 o 
 
 
 
 
 
 Cd 
 
 r; 
 
 
 1 
 
 
 z < 
 
 
 t*S 
 
 o 
 
 
 
 h-t 
 
 
 
 C3 
 
 o 
 
 g 
 
 
 
 w 
 
 
 
 
 
 
 O 
 
 w 
 
 En 
 
 En 
 
 B 8: 
 
 Z 
 
 > 
 
 
 
 s 
 
 
 
 
 ft O 
 
 
 
 Z 
 
 Is 
 
 0. 
 
 Z 
 
 
 
 H 
 
 w 
 
 
 
 
 W o 
 
 
 W 
 
 « 
 
 
 w 
 
 O 
 
 n 
 
 J_ p £ fa 
 
 
 a 
 
 
 0, 
 
 
 h 
 
 
 e- O 
 
 
 c a 
 
 W 
 
 
 n x 
 
 M 
 
 
 O 
 
 u 
 
 w 
 
 £ 
 
 
 
 5 
 
 
 > 
 
 
 3o 
 
 c 
 
 < 
 
 (S w w 
 
 
 (4 
 
 P. 
 
 § 
 
 ►J 
 
 o 
 
 B 
 
 
 
 o 
 
 o 
 
 CC 
 
 Eh 
 
 o e d 
 
 g 
 
 J 
 
 
 
 w 
 
 
 >H 
 
 
 z o 
 
 z 
 
 o 
 
 
 w a 
 
 a, 
 
 
 
 
 o 
 
 Eh 
 
 M 
 
 >i 
 
 < 
 
 
 >n Cd 
 
 
 
 a 
 
 O < K. 
 
 fr- 
 
 
 § 
 
 nK WO 
 
 CO 
 
 £ 
 
 
 
 z 
 
 .4 
 
 w 
 
 * .J 
 
 Ei 
 
 o: m 
 
 J 
 
 
 M 1 
 
 ee 
 
 C-. 
 
 J «r E- ,J 
 
 c c c-> <: 
 
 
 
 
 o 
 
 a, 
 
 s 
 
 O J 
 
 t-i 
 
 ti E- 
 
 < 
 
 o 
 
 E-i M C- 
 
 >-" 
 
 1 
 
 
 £-. 
 
 z 
 
 J 
 
 u; 
 
 a 
 
 CO 
 
 K O 
 
 Q 
 
 O < 
 
 
 w 
 
 S ~ S 
 
 00 
 
 En 
 
 z X >- M 
 
 g 
 
 o 
 
 e 
 
 En 
 
 :- 
 
 H 
 
 H 
 
 = C 
 
 W 
 
 (X ft 
 
 o 
 
 K 
 
 £ >n S 
 
 1 
 
 
 < 
 
 KP.KD 
 
 CJ 
 
 CO 
 
 CO 
 
 a 
 
 E- 
 
 O Eh 
 
 
 A- W 
 
 S3 
 
 h- ( 
 
 En 
 
 w 
 
 E- 
 
 S5S 
 
 w 
 
 
 
 
 z 
 
 
 
 
 
 
 
 
 
 P. 
 
 s 
 
 3 < o 
 
 CO 
 
 
 
 
 o 
 
 
 
 
 
 
 
 
 13 
 
 c- > 
 
 o 
 
 < s <-> 
 
 X 3 
 
 « 
 
 
 JO o 
 
 o 
 
 d 
 
 .a 
 
 U 
 
 ■o 
 
 01 
 
 
 t~ 
 
 w 
 
 CO rH 
 
 1 i 
 
 m cvj 
 
 © >- O 
 
 
 
 41 41 O 
 
 
 
 CC *> 
 
 
 
 C TJ 
 
 
 
 '1 '.' 11 
 
 
 
 coo 
 
 
 
 •Z3 +> 3 
 
 m 
 
 
 rH c > 
 
 O 
 
 -s 
 
 ••-1 D 13 
 
 C 
 
 t— 
 
 3 e < 
 
 
 ON 
 
 ,0 D. 
 
 rH 
 
 rH 
 
 P. Ih 
 
 iH 
 
 
 ^n m 
 
 M 
 
 
 x: u v, 
 
 
 u 
 
 to > 
 
 
 41 
 
 u ^ 
 
 « 
 
 ^ 
 
 r-j Q 4? 
 
 s 
 
 a 
 
 41 
 
 > T3 C 
 
 £1 
 
 > 
 
 a! q o 
 C 3 o 
 
 h 
 
 O 
 
 3 
 
 
 e a 
 
 < o 
 
 z w 
 
 *n (L, 
 
 O W 
 
 O H 
 
 -325 
 
 ►n 
 J 
 
 9 
 
 < 
 
 r- 
 
 a g a 
 
 si* 
 
 tH ft 
 
 w 
 
 fn O 
 
 ~. M 
 
 'i [h 
 
 o <: 
 
 u. ^: 
 
 6: E- 
 
 o p 
 
 u n 
 
 Eh U 
 
 K i.i 
 
 CO 0, 
 
 M Eh 3 
 
CERL HDB Recommendations 
 
 27 
 
 a 
 
 a 
 
 s a 
 
 o u 
 o -d 
 
 n oh x 
 
 c 
 v 
 u 
 a 
 p. 
 
 -p 
 
 C 
 
 >~ 
 
 +> 
 
 
 
 -v-i 
 
 'J 
 
 £ 
 
 tr 
 
 
 
 > 
 
 ■3 
 
 Pc 
 
 
 
 
 
 II 
 
 
 •O 
 
 
 
 
 O 
 
 
 
 
 
 ■g 
 
 ■J 
 
 BD 
 
 
 
 
 4. 
 
 t, 
 
 II 
 
 
 
 o 
 
 
 
 >H 
 
 »>-■ 
 
 tc 
 
 B 
 
 (B 
 
 Q 
 
 is 
 
 83 
 
 3g 
 
 
 | 
 
 
 >-> 
 
 C 
 
 
 
 ► 3 w 
 < to 
 
 ^ 
 
 p 
 
 M 
 
 6 
 
 s 
 
 S5k 
 
 
 CO 
 
 v> 
 
 n- 
 
 »l' l: 2= 
 
 M 
 
 a f> o « •> %i 
 
28 
 
 CLPL HD& Recommendations 
 
 m n to co 
 
 O V. 
 
 <U ,-! ^ 
 
 .-I ,c .o 
 
 * c cc 
 
 a >- i.. 
 
 P ^ -i 
 
 r-J f. W 
 
 d O 1, 
 
 > & a 
 
 "3 "3 
 > > 
 
 S5 
 O 
 
 ?5 < 
 
 o o 
 
 w EH 
 w P 
 
 Q O 
 W 
 
 a w 
 
 I u 
 
 h ° 
 
 O £- 
 
 O < 
 
 cc o 
 
 ft* HH 
 
 ft. > 
 
 Q < 
 o < 
 
 o 
 
 sS 
 
 E-..E-. 
 
 a. z 
 o w 
 
 c a 
 
 "I 
 
 x eh 
 o s 
 < o 
 
 o >-. ^ 
 
 o -3 
 
 W ft. I o 
 
 .-; =2 
 
 d. >- Uj h 
 
 >-< z »-• cc 
 
 *-. < ^ S 
 
 t; - ° £ 
 
 2 H 
 w o 
 
 §3 
 
 P h5 
 
 o 5 
 
 ft- * 
 
 CO CC 
 
 3 
 
 w o 
 
 w o 
 
 = K 
 
 O E-. 
 
 flip Cj13 O 
 
 < i so 
 
 
 w < o t< o 
 
 
 O 
 
 B 
 
 M W 
 
 E-t O 
 
 m <: 
 
 £ Sg 
 
 E- 
 
 _) 
 
 i- 
 
 -J 
 
CF.RL HDB Recommendations 
 
 ?9 
 
 to t» 
 u 
 
 a t> 
 
 X .4- 
 
 8 B 
 
 tD D 
 01 O 
 
 -a 
 
 f> 
 
 
 <a 
 
 
 43 
 
 
 a 
 
 
 
 § 
 
 U 
 
 >8 
 
 
 u 
 
 n 
 
 m 
 
 •^ 
 
 •^ 
 
 a> 
 
 •1 
 
 > 
 
 EC 
 
 -^ 
 
 3 
 
 +i 
 
 
 3 
 U 
 
 a 
 
 a> 
 
 o 
 
 a> 
 
 
 c 
 
 
 ft 
 
 u 
 
 ■" 
 
 "8 
 
 h I, «. 
 
 •a -o -a 
 «! !>. >. !> 
 
 ,0 <-> <J *> 
 
 i r i; < 
 
 r-J ,-. -1 -H 
 ffl r-4 r-l #H 
 
 
 o 
 
 5Q 
 
 28 
 
 a 
 
 s ° 
 
 eg 
 
 83 
 
 
 sS 
 
 
 s 
 
 S= 
 
 t- 
 
 »-( 
 
 O 
 
 2 
 
 f- o 
 
 HI 
 
 
 t- 
 
 t- 
 
 ill 
 
 K to n 
 
 3 
 
 re 
 
 fa 
 
 g E 
 
 o o 
 
 IS i? a 
 
 SGo8 
 e~ w a. 5 
 
 ifltOOO 
 
 <0 X) C "O l> Vh 
 
 sjh 
 
 I-; U ft. 
 
 A. <•. S 
 
 E >- J 
 
 ►-. < -5 
 
 jJh 
 
 t a. b 
 
 m a. 
 
 o o CO 
 
 <d ,o «j 
 
 o C ac 
 
 5gS 
 
 to CJ 
 
 aJ a~- '- -O 
 
30 
 
 CERL HDB Recommendations 
 
 This is an important question laroely because of the 
 research which tends to cast doubts on the consistency of 
 manually prepared document analysis. Salton reports in a number 
 of research studies that automated lanquaqe analysis procedures 
 can provide benefits. One of the major results of a recent study 
 was that although simple word extraction followed by boolean 
 search does not produce retrieval results equivalent in 
 effectiveness to standard manual indexino techniques/ a variety 
 of techniques can be added to obtain retrieval whose 
 effectiveness exceeds conventional manual methodologies. When 
 these factors are added to the expense of preparing the index and 
 thesauri by hand/ the argument to stay with automated techniques 
 becomes stronger. 
 
 One wonders openly whether the choice to implement the HDB 
 in the way it now appears was made with full understanding of the 
 implications of this ever-expanding body of research or merely as 
 a result of the convenience/ or even the personal bias of one of 
 the early workers on the project. Certainly the CERL HDB 
 managers have a choice between two courses of action. One is to 
 use a manual indexing technique coupled with a manually prepared 
 thesaurus or set of key words. Given this choice several of the 
 commercially available and tested retrieval systems could be 
 adapted and the HDB would be searched with techniques similar to 
 those now successfully being used to search the major document 
 data bases in use today (NTIS/ ERIC/ Che-n Abstracts/ etc). The 
 
CERL HDB Recommendations 
 
 31 
 
 other choice is to continue with the more forward-looking but 
 less proven techniques of automatic content analysis with natural 
 language queries as represented by the SMART system. The system 
 in use today by the HDB lies somewhere in between the two 
 extremes/ since laborious indexing is done as the statements are 
 prepared/- and boolean searches of a sort are done on the basis of 
 these indexes. But this is complemented by running SWART, which 
 does not ma"ke use of the indexes at all. 
 
 6.2 Suitability of SMART 
 
 There are some questions and reservations about the use of 
 SMART as a major tool for the HDB. 
 
 First of all, the SMART implementation requires a very large 
 core region on a 360 system. The current implementation was 
 intended as a vechicle for experimental work in information 
 retrieval techniques. As a result/ much of the size of the code 
 is concerned with measuring retrieval performance. Considerable 
 portions of the code which is loaded from the SMART library is 
 never actually executed. A production version could conceivably 
 be produced which would not include as many measurinu tools and 
 thus could be somewhat smaller. One current goal of the Cornell 
 group is a modular implementation so that one could load only 
 necessary modules for a production environment implementation. An 
 alternate solution would be to implement the code (which is 
 primarily Fortran with some assembly language subroutines) on a 
 virtual memory operation system. 
 
3? 
 
 CERL HDP Recommendations 
 
 Secondly* the current implementation is strictly a batch 
 system. An attempt has been made at Cornell to implement SMART 
 under TSO* but that work now appears to have fallen by the 
 wayside. It seems apparent that an interactive system would make 
 it easier for the user to modify his searching strategy based 
 upon what he is finding* rather than submitting a number of batch 
 jobs* all of which must do the complete search. As an 
 information system to be used by information specialists this 
 major inconvenience miaht be overcome* but in our view it is 
 unlikely to ever be regularly used by customers directly in this 
 mode . 
 
 Some thouaht should be given to why commercially available 
 systems are not offering automatic content analysis and natural 
 language queries in quite the same way that SMART attempts to do. 
 The systems which are available commercially seem to be 
 universally built on some variation of key word searching and 
 boolean expressions for search requests. The commercial systems 
 have a long (up to 10 years) period of development behind them. 
 When these efforts started natural language processing was not 
 sufficiently developed to make it worth the commercial risk. 
 Some would argue that it is still not worth the risk. The fact 
 that so many commercial systems use key words tends to suggest 
 that the technology is accepted and a long term period of support 
 can be envisioned. The implication of all of this to the HD3 is 
 that if what is needed must feature natural lanauage queries with 
 
CERL HDB Recommendations 
 
 33 
 
 no manual preparation of key words* a non-commercial* semi- 
 experimental system is the only choice. However* if the current 
 HDB can be expanded (either by hand or with proorams) to include 
 key words* one of the commercially available systems will provide 
 reliable long term service of a less sophisticated nature. It 
 may even be that simple full text searchinq of the statements 
 themselves using a controlled thesaurus* could be used on one of 
 the commercial systems. 
 
 These conclusions should not preclude pursuing the i o a I of 
 interactive languaqe query systems. To whatever extent this 
 capability is crucial to the lonq term goals of the COF* it 
 should be pursued as an adjunct to systems like the HDB. 
 However* a completely adequate job of data storaae and retrieval 
 in support of an Information Access Center (1AC) for the HDB can 
 be done with commercially available systems. Unfortunately* some 
 backtracking will be necessary to associate appropriate keywords 
 with each of the habitability statements if that course is taken. 
 
 6.3 The Choice of Continuing *J i t h SMART 
 
 One possible course of action is to continue usino the SMART 
 system in its present form. This can be done with or without the 
 concurrent use of the package of programs loosely associated with 
 AND/OR. Options that are directly available to CERL at the 
 present tiree include continuing with the CSO installation and 
 running the software that is now available* transferring the 
 SMART system to the Amdahl installation at the Univprsity of 
 
34 
 
 CFRL HDB Recommendations 
 
 Michigan/ or transferring it to the IBM 360/91 at UCLA. Of 
 course/ it is always possible to put the system on some 
 nationally available time-sharing service/ but that would cause 
 some (solvable) difficulty with the local availability of 
 printouts. 
 
 The DEC-10 and 360/75 installation at the Computing Services 
 Office (CSO) of the University of Illinois is expected to stay 
 available in its present form only through the middle of 1977. 
 At that time the present indication is that the DEC-10 system 
 will be taken out of service. The general expectation is that 
 the 360 will stay in service through the middle of 1978/ because 
 of the demand by university users for whom conversion will be 
 impossible before that time. Thus/ there need be no rush to 
 bring up a different system if one is willing to tolerate the 
 Ion a turn-around time for SMART jobs. Some of this turn- a round 
 time is a result of having to ask that the disk with the HOB be 
 mounted each time it is needed. Requestinq that the disk be 
 permanently mounted would reduce that delay but produce some 
 small increase in costs. 
 
 Another course of action/ if the choice is to stay with the 
 SMART system/ is to move the library to the installation at the 
 University of Michigan at Ann Arbor. At least two projects at 
 CfcRL are currently usina the University of Michigan system with 
 apparently good results. If the choice is the Michigan system/ 
 then the next choice is what to do about the AND and OR proarams. 
 
CERL HDB Recommendations 
 
 35 
 
 However/' the very nature of the Michigan Time Sharinq (MTS) 
 system provides a useful solution. MTS is a superior time 
 sharinq system which allows interactive access from user programs 
 in a rather general way. The system has an impressive collection 
 of interactive services/ including a oood editor/ document 
 preparation systems/ and convenient handling of large disk files. 
 It would be possible to recode the AND and OR proorams/ as well 
 as the programs which allow one to see the documents (statements) 
 and bibliographic entries. The programming could all be done 
 from CERL with interactive terminals. Terminal access to tt T S can 
 be directly by FTS line/ or can be arranged in the same way that 
 some other projects at CERL employ. They dial to a phone port at 
 CAC which is attached to a multiplexer/ the other end of which is 
 a port on the MTS system. 
 
 The multiplexer equipment now in use is available as excess 
 capacity on a system installed by CAC for another project. That 
 project is currently expected to continue at least throuoh 
 January of 1977. The excess capacity of this line is expected to 
 be available so long as that project continues to be funded/ 
 which is expected to be for more than another year. In the worst 
 case/ that in which the CAC project no longer needs the access to 
 MTS/ it would only require four regular users at CfcRL to justify 
 pooling costs to put in this equipment themselves/ pay the same 
 rate each that is currently being uaid by CERL users tor this 
 service/ and have the multiplexer strictly for CERL use. The 
 
36 
 
 CERL HOB Recommendat ions 
 
 total budget for the multiplexer connection is on the order of 
 $80 (J per month. As few as four projects could use such a 
 communications system to keep their total costs below long 
 distance access costs. 
 
 Remote job entry from the Unix system at CAC is now 
 available. Files of card i^aoes are transmitted to ^TS from Unix 
 disk files in a manner similar to Hasp work stations. Printed 
 output from UM is available on the equipment at CAC. This 
 service is expected to continue so long as the line to Michigan 
 is needed and there are funds to support it. Charaes for use of 
 the local system come as a separate bill from the computer 
 charges assessed at Michigan. The communications cost is 
 currently billed as a fixed monthly cost for the use of the 
 multiplexer and associated phones. 
 
 The costs at UM are said to be reasonable accordinq to the 
 CAC users of MTS . The only noticeable startup costs for going to 
 this solution would be costs associated with sendinr the HDB 
 files to Michigan^ and the costs of reprog r ammi rig the programs 
 other than SMART which are needed to continue the present mode of 
 operation. However/ our experience with *TS indicates that the 
 level and reliability of the service at Michigan warrant its 
 serious consideration. 
 
CERL HDB Recommendations 
 
 37 
 
 Still another avenue is open to facilitate stayinq with the 
 SMART system without being concerned with the continuing 
 availability of the CSO 360. The Campus Computing Network of UCLA 
 is available on the ARPANET and can be accessed as easily from 
 any arpanet node as it can from the CAC. The time sharina system 
 there is TSG/ which certainly does not comcare to f^TS in terms of 
 its friendliness to the user. However/ larae batch jobs can be 
 run at CCN/ and printouts can be returned to the printer at CAC. 
 The charges for this connection could probably be kept to on the 
 order of $!> per hour connected to the network/ plus the normal 
 user fees at CCfo. The 360/91 installation at tCN is one of the 
 more reliable places we have come in contact with over the last 
 two years. 
 
 The interactive portions of the HDB tasks would have to be 
 receded under the TSO system at CCN. However/ since they ere now 
 coded in Fortran/ a mere conversion would suffice to make the 
 system as useable in that environment oS it is in its present 
 environment. Costs there are comparable to costs at the 
 University of Illinois/ except that in our experience jobs which 
 require a large region size (as SMART does) generally are cheaper 
 to run at CCN. Also/ because of more core on the CCN system/ 
 large jobs can be run at any time of day and the turn-around time 
 is generally better than for a comparable large job on the 360/75 
 at CSO. Also/ the processor out there is much faster and as a 
 result/ the wait for results of a query should be much shorter. 
 
38 
 
 CERL MD8 Recommendations 
 
 In either the University of Michiaan or the UCLA situations* 
 the disadvantage of SMART being a strictly batch system would 
 still apply. However/' with the appropriate cooperation of the 
 originators of SMART at Cornell/ either of these systems would be 
 suitable for converting SMART into an interactive system. This 
 would be no small undertaking. It could not (or should not) be 
 done without the active cooperation of the group at Cornell who 
 are intimately familiar with the inner workings of SMART. The 
 software development for such a task would conservatively take 
 about a year for about a two to two and one-half man-years of 
 prograreroi nq . 
 
 As a purely batch system* SMART could be installed on one of 
 the nationally available time-sharing systems which offers IBM 
 equipment. If the remote job entry equipment at CERL is sometime 
 attached to such a service* it would be easy to move a copy of 
 the SMART library to such a service and run just the SMART system 
 as pure batch jobs. If the system also supports time-sharing 
 service* the AND and OR programs could be recoded just as they 
 would have to be with any of the other choices. 
 
 7. Developing a Replacement for SMART 
 
 While the SMART system in its present implementation is not 
 quite satisfactory for the production stages of the hDB effort* 
 careful consideration must he given to any proposals to change 
 systems at this staoe of development. Certainly a charge from 
 the implementation on the IBM 360 and DtC-10 system at CSO is 
 
CERL HDB Recommendations 
 
 39 
 
 going to be necessary/ because those systems are scheduled to be 
 phased out of service over the next two years. Section 6 
 discussed some of the issues which must be addressed in an 
 information retrieval system suitable for the HDB* but outlined 
 the options available to CERL if their decision were to stay with 
 the SMART program. 
 
 The assumption in this section is that a decision has been 
 made to abandon the SWART proorams and develop or find somethinn 
 else. 6iven that assumption* two avenues of investigation are 
 open. One is to develop the CELDS system which is performing a 
 similar function for environmental data bases in conjunction with 
 another group at CERL. The other is to make the necessary 
 modifications to the HDB to make the information retrievable 
 using one of the nationally available information retrieval. The 
 DIALOG system at Lockheed is given as an example because it comes 
 the closest to meeting the criteria outlined in Section 5. 
 
 7.1 Revision of CELDS for the HDB 
 
 Any initial implementation of HDB on a CELDS-copy retriever 
 would have to include at least the capabilities that ANDOR* BIBAX 
 and DOCAX already provide to HDB users. CELDS provides these 
 options now* and in addition provides: 
 
 1) all functions are combined into one retrieval language. 
 
 2) SAVE interesting and often used output sets 
 
40 
 
 Cl-RL HDB Recommendations 
 
 3) HELP 
 
 4) partial search tells user how many statements satisfy 
 
 sub-expressi ons 
 
 5) parentheses and full expression nest inn 
 
 6) OOPS to return to previous statement-set 
 
 7) allows multiple values per field 
 
 8) off-line printing 
 
 9) simple logon-louoff 
 
 10)retains fast response time even for very larae databases 
 
 To convert to a field-oriented system (like CELDS) the HDB 
 
 could be broken into the following fields: 
 
 ACC - accession number 
 
 DOC - document number 
 
 STMT- statement number 
 
 DATE- date published/ researched/ input 
 
 [unknown for current database! 
 BIB - bibliographic data 
 
 AUTH- name of author(s) [unknown for current DB3 
 FUNC- functional area code 
 TRFC- training facility code 
 PHYS- physical settings 
 
 ENV - environmental descriptors (however many apply) 
 0CCU- occupants 
 PSTR- posture 
 INVM- involvement 
 0RGF- organizational functions 
 SFCN- function of statement 
 TEXT- the text of the statement 
 KEY - keywords Cunknown for the current DB3 
 
 Several new values would have to be added to the SFCN field 
 
 including "objectives"* "data"* and "procedure". Several of the 
 
 fields (such as PSTR and 1NVM) could be dropped and their values 
 
 used as KEYWORDS. It would help streamline the list of fields 
 
 without loss of generality. The DATE field is a useful field to 
 
CERL HD3 Recommendations 
 
 41 
 
 include* but not strictly necessary. The only non-searchable 
 fields would be BIB* TEXT/ and DATE. 
 
 CELDS-like format includes one line per field and each line 
 is prefixed by accession number and field number. The current 
 HOB lines are suffixed by statement number/ card number/ and 
 document number* and separated (unnecessarily) by 'NEXT TEXT 1 
 cards. Converting data formats would be fairly simple/ except 
 that a few desirable fields would be missing I e.g. keywords] and 
 the current HDB uses dioit strings for the indexes. Names would 
 be much easier for novice users to read. These could be 
 converted automatically. 
 
 Two CELDS input proaratis would have to be modified sliohtly 
 (made more general) to accomodate the different field names. The 
 CELDS retriever program would have to be modified to use the new 
 fields also. The inversion program woulo have to be run on the 
 newly created Habitability Data Base. 
 
 The next obvious improvements would include adding keywords 
 to the database/ and adding an on-line thesaurus to the 
 retriever. Then the combined retriever could be modified to use 
 the thesaurus to recognize concepts in a very SMART-like 
 environment. Concept numbers and weighting are not currently 
 practical for interactive searching/ but this could make a 
 fascinating research project. 
 
 7.2 Using a Commercial Information Retrieval System 
 
 One of the more popular and widely used of the commercially 
 
42 
 
 CLRL HDB Recommendations 
 
 available information retrieval systems is the DIALOG system 
 operated by Lockheed in Palo Alto* California. This system was 
 included in the survey discussed in Section 5. If a commercially 
 available system is considered as a home for the HDR/ certainly 
 DIALOG should be considered a prime candidate. 
 
 The decision to move to a commercial retrieval system 
 presents questions both of a technical nature and of a purely 
 operational nature. We address both kinds of questions/ but from 
 the very limited basis of the specific information which is 
 available to us in the course of this investigation. We consider 
 first the technical questions of what would be reouired to put 
 the HD3 into the DIALOG system. 
 
 Putting the HDB into DIALOG would require almost exactly the 
 same amount of effort as putting it into CEL DS-f ormat . DIALOG is 
 a field -oriented system with full text searchinq ability* but not 
 natural language query. The HDB would almost certainly have to 
 be converted to a DIALOG format* and keywords should be added. 
 DIALOG would require a very complete thesaurus* which would then 
 be available on-line. Full text searching in DIALOG requires an 
 exact match to the words in the statement. 
 
 The DIALOG system works primarily with searches on 
 predefined fields. Although the system is designed for 
 b ibl i oqraphi c retrievals the similarity to the information in the 
 HDB suggests that only a small perturbation of the H D ^ would be 
 required for conversion to DIALOB. The fields in the index used 
 
CERL HDB Recommendat i ons 
 
 A3 
 
 with the HDB statements could be made into fields in the DIALOG 
 sense. The statements in the HDD are similar to abstracts and 
 thus could be treated by DIALOG in the same way abstracts are 
 treated. The task of convertino what now exists in HDB to a form 
 suitable for DIALOG could be assisted by some of the text 
 processing capability in the UNIX system at CAC. 
 
 One conceptual dissimilarity between the two systems is that 
 in DIALOG all information of one record (or set of records) 
 concerns a sinqle document/ and there is no field to refer to a 
 parent document. In HDB/ on the other hand/ the basic 
 information is a statement/ several of which come from the same 
 parent document. It would be possible to think of each HDB 
 statement as a document in the DIALOG sense/ providing that an 
 extra field is added to yi v e reference to the parent document. 
 Also/ in this context/ it would probably be advisable to encode 
 keywords for each of the Hu3 statements. other information would 
 be based on the parent document. This would probably need to 
 include the author or some other reference to the source/ the 
 date if that applies/ the corporate author if one exists/ and 
 inevitably/ the key words for the parent document. 
 
 Another pressing need/ in the event of this choice as well 
 as several others/ is for a completed thesaurus for the HDB. The 
 approach taken in the thesaurus for the early rortions of the HDB 
 is a step in the right direction/ but it needs to be expanded to 
 include terms peculiar to the whole ranqe of habitability 
 
44 
 
 CLPL HDB Recommendations 
 
 statements/ not just the limited subset available to CERL now. 
 By standardizing the HDR vocabulary* and by carefully keywording/ 
 DIALOG could be a fast/ easy to use system for retrieving from 
 the HOB. 
 
 It is impossible to determine the cost of puttina the HDB on 
 DIALOG nor estimating what it would cost to run/ except by 
 comparing the complexity of HDB to some of the other available 
 databases for which at least representative user costs are 
 available. The cost for accessing the NTIS data base/ for 
 example/ is SZb per connect hour. (The system is purely 
 interactive.) In addition to this a communication cost is added 
 dependent upon the mode of access. For access via Telenet this 
 charge is $8 per hour. Since the HDB is considerably smaller 
 than NTIS/ one would expect the charge to be less/ except for the 
 fact that fewer customers might mean hi q her prices. 
 
 The documentation of DIALOG makes it \/ery clear that they 
 will not be able to predict the cost for a new database snd the 
 accompanying service to access it. Such an estimate could be 
 nothing but a raw guess without an extremely detailed proposal 
 from Lockheed. One immediate suggestion is that Lockheed should 
 be contacted/ given as much information as possible about the 
 HDB/ includinq this report/ and then asked to submit a cost 
 proposal. Sales brochures for DIALOG indicate the price for out 
 of the ordinary services as "negotiable." 
 
CERL HOB Recommendations 
 
 45 
 
 Although we have no idea what it costs to put up the NTIS 
 information on DIALOG* it seems clear that one of the 
 justifications is the wide interest in accessing the NTIS 
 database* and thus the customer base with which to recover the 
 installation costs. For a special purpose client like CERL the 
 cost cannot be spread over so many customers and thus the 
 apparent cost will seem higher. In order to operate an IAC which 
 includes the capability to search the HDB for clients* CERL would 
 thus have to pass on fairly high operational costs to the client 
 or operate the service at a loss until the number of clients 
 spreads the cost out over a wider base of users. 
 
 There is another whole question which is still unanswered as 
 to whether Lockheed would even be interested in putting HDB on 
 their system. Certainly the customer base at the present time 
 would not warrant their covering the cost of transformino the HDB 
 into a form suitable for DIALOG. CERL would either have to do 
 that themselves or pay Lockheed to do it. Now it is certainty 
 true that Lockheed is intended to be a profit making venture* and 
 thus they may be willing to put whatever someone wants onto their 
 system for an appropriately larqe sum of money. However/ it may 
 be that their growth plans do not allow for yet another 
 potentially large data base to come on the scene in the near 
 future. If this is true they will not be able to put the HDB 
 database on DIALOG* regardless of whether or not they could 
 recover their costs for doing so. We were able to contact DIALOG 
 
46 
 
 CbPL HDB Recommendations 
 
 users, and use DIALOG on-line. The DIALOG users we sailed were 
 largely pleased with Lockheed service. 
 
CERL HDB Recommendations 
 
 47 
 
 LIST OF REFERENCES 
 
 [13 T. A. Davis* "Conceptualization of Habitahility Expressions 
 
 Habitability*" Data Base U.S. Ar.-ry Construction 
 
 Research Laboratory Interim Report D - 6 b * U.S. 
 
 Engineering Research Laboratory a 
 
 for the 
 
 Eng ineer i ng 
 
 Army Construction 
 
 Champaiqn* Illinois* Auqust 1976 
 
 123 Dennis W. Fife* Kirk Rankin* Elizabeth Fong* Justin C. 
 Walker and Beatrice A. Barron* "The Technical Index of 
 Interactive Information Systems*" U.S. Department of 
 Commerce National Bureau of Standards Technical Kote No. 
 819* U.S. Government Printing Office/ Washington/ 1974 
 
 m 
 
 Wayne Hamilton* "General Overview of the 
 Portion of an unpublished report* 1976 
 
 HDB 
 
 File System*' 
 
 [43 U.S. Army Construction Engineering Research Laboratory* 
 "Demonstration Instructions for the Prototype Habitability 
 Data Base*" U.S. Army Construction Engineering Research 
 Laboratory* Champaion* Illinois* ( rt i meoi raphed ) 
 
 [53 U.S. Army Construction Engineering Research Laboratory* 
 "User's Instruction Manual for the Prototype Habitability 
 Data Base*" U.S. Army Construction Engineering Pesearch 
 Laboratory/ Champaign* Illinois* (Mimeographed) 
 
 [63 Gerard Salton* ed./ The SMART Retrieval System* 
 Hall* Inc.* Enqlewood Cliffs* New Jersey* 1971 
 
 Prentice- 
 
48 
 
 CtRL MDB Recommendations 
 
 BIBLIOGRAPHY 
 
 [13 G. J. Baker/- "Database world/-" Database Journal/ Vol. 6/ No. 
 11/ 1976/ pp. 18-2 3 
 
 [23 G. J. Baker/ "The Correct Use of Codasyl DBTb Sets/" 
 Database/ Vol. 6/ No. 2, pp. 19-21 
 
 [33 Barnett and Lightfoot, "Information Management Systems 
 (IMS)/ a User's Experience with Evolutionary Development/" 
 in Data Base Management Systems/ Proceedings of the SHARE 
 Working Conference on Data Base Management Systems/ 
 Montreal/ Canada/ July 23-27, 1973/ Donald A. Jardine, ed./ 
 North Holland Publishing Company/ 1974 
 
 £43 M. Bibby/ "User Experience with IDS at STC/ M Database 
 Journal/ Vol. 6/ No. 5/ 1976/ pp. 7-13 
 
 L53 Vaclav Chvalovsky/ "Anything New in Data Base Technology?/" 
 Datamation/ Vol. 22/ No. 4/ April 1976/ pp. 54-55 
 
 C63 L. J. Cohen/ "Is Database the Way of the Future? 
 Journal/ Vol. 6/ No. 11/ 1976/ pp. 15-18 
 
 Database 
 
 [73 Leo J. Cohen/ Data Base Management 
 Development Corporation and Q. E. D 
 Inc./ 1973 
 
 Systems/ Performance 
 Information Sciences/ 
 
 [83 Robert M. Curtice/ "The Outlook for Data Base Manaoement/" 
 Datamation/ Vol. 22 '/ ho. 4/ April 1976/ pp. 46-49 
 
 [93 C. J. Date/ "Relational Data Base Concepts/" Datamation/ 
 Vol. 22^ No. 4/ April 1976/ pp. 50-53 
 
 CIO] T. A. Davis/ "Conceptualization of habitability Expressions 
 
 for the Habitability/" Data Base U.S. Army Construction 
 
 Engineering Research Laboratory Interim Report D-68/ U.S. 
 
 Army Construction Enaineerinn Research Laboratory/ 
 Champaign/ Illinois/ August 1976 
 
CERL HOB Recommendations 
 
 49 
 
 [11] E. J. Emerson/- *DW-S 11/U r User Experience/' in Data Base 
 Management Systems/ Proceedings of trie SHARE forking 
 Conference on Data a a s e Manaqement Systems/ Montreal/ 
 Canada/ Julv 25-2 7 / 1973/ Donald A . J a r d i n e / e d . / North 
 Holland Publishing Company/ 1974 
 
 [123 Mrs. S. Fenlon/ "The On-Line Patient Index at Addent r ook e * s 
 Hospital/ Cambridge/" Database Journal/ Vol. 6/ No. 7, 1976/ 
 pp. 22-26 
 
 [133 Dennis W. Fife/ Kirk Rankin/ Elizabeth Pong/ Justin C. 
 Walker and Beatrice A. M a r r o n / "The Technical Index of 
 Interactive Information Systems/" U.S. Department of 
 Commerce National Bureau of Standards Technical Note No. 
 819, U.S. Government Printinq Office/ washinoton/ 19 74 
 
 [143 Wayne Hamilton/ "General Overview of the H D B file 
 Portion of an unpublished report/ 197 6 
 
 Sy st em/ 
 
 [153 D. Hannaford/ "TOTAL--A Detailed Analysis/ 
 Journal/ Vol. 6/ No. 7, 1976/ pp. 9-14 
 
 Database 
 
 [163 L. C. Hobbs/ "Future Trends in Hardware/" in ^ata Base 
 Management Systems/ Proceedings of the SHARE Working 
 Conference on Data Base Management Systems/ Montreal/ 
 Canada/ July 2 3- 2 f / 1 9 7 3 / Donald A . Jardine# e d . / n o r t h 
 Holland Publishing Company/ 1 Q 74 
 
 [173 Susanne M. Humphrey/ "Searching the i'iEDLAPS Citation File 
 On-Line Usin- ELM ILL and STAIRS: " Updated Comparison/" 
 Information Processing. K ^anafpneit / Vol. 1 ? , Perqamon 
 Press/ Great Britain/ 19 76/ pp. tl-70 
 
 [183 F. E. Johnson/ "IDS — A Brick in the Database Tower of 
 Babel/" Database Journal/ Vol. 6/ No. b/ 1976/ pp. 2-6 
 
 [193 Database Journal/ "The Interim Report of the ANSIX3SPARC 
 Study Group on Database Manaaement Systems/" Database 
 Journal/ Vol. 6/ No. 11/ 19?6/ op. 10-14 
 
 [203 Database Journal/ "The IBM Database Ranne--V ANDL I / DLI 
 
5U 
 
 CERL HDB Recommendations 
 
 ENTRY/ DLI and IMS/ 
 1976/ pp. 2-9 
 
 Database Journal/ Vol.6/ No. 10/ 
 
 f.213 Database Journal/ "Database World/" Database Journal/ Vol 
 6/ No. 10/ 1976/ or. 19-23 
 
 [223 Database Journal/ Vol. 6/ No. 5/ "IDMS and the 2900 Series/ 
 Database Journal/ Vol. 6/ No. 5/ 1976/ p. 19 
 
 [233 Donald W. King and Edward C. Bryant/ The Evaluation of 
 iDiormation Services and Products/ Information Resources 
 Press/ Washington/ 1 9 H 
 
 [243 U.S. Army Construction Engineering Research Laboratory/ 
 "Demonstration Instructions for the Prototype Habitability 
 Data Base/" U.S. Army Construction Enoineering Research 
 Laboratory/ Champaiqn/ Illinois/ (Mimeographed) 
 
 C253 U.S. Army Construction Engineering Research Laboratory/ 
 "User's Instruction Manual for the Prototype Habitaoility 
 Data Base/" U.S. Army Construction tnqineerinq Research 
 Laboratory/ Champaign/ Illinois/ (Mimeographed) 
 
 [263 P. A. Lavallee and S. Ohayon/ "DMS Applications and 
 Experience/" in Data B3se Management Systems/ Proceedings of 
 the SHARE Working Conference on Data Base Management 
 Systems/ Montreal/ Canada/ July 23-27/ 1973/ Donald A. 
 Jardine/ ed./ North Holland Publishing Company/ 1974 
 
 [273 E. T. Lee/ "Shape-Oriented Stora-ie and Retrieval of 
 Geometric Figures and Chromosome Images/" Information 
 Processing & Management/ Vol. 12/ p e r g a m o n Press/ Great 
 Britain/ 1976/ pp. 35-41 
 
 C283 S. Lie-Nielsen and J. Refsnes/ "SIBAS — The Portable Database 
 Management System/" Database Journal/ Vol. 6/ No. 11/ 1976/ 
 PP. 2-9 
 
 [293 J. Lockeret2/ "User Experience with IMS 
 Company/" Database Journal/ Vol. 6/ No. 
 
 at tsso Pet ro leum 
 10/ 1976/ pp. 10-18 
 
CERL HDB Recommendations 
 
 51 
 
 C30J I. G. MacDonald/ "Univac's Interpretation of the CODASYL 
 D87G Proposals/" Database/ Vol. 6/ No. 2 , pp. 3-7 
 
 [31] R. Naskell/ "LEXICON--An tstablished Data Dictionary 
 System/" Database Journal/ Vol. o / No. 7 / 1976/ pp. 15-21 
 
 [323 W. E. Mercer/ "User Experience •TOTAL*/" in Data Base 
 Management Systems/ Prnceedinas of the SHARE Working 
 Conference on Data base Management Systems/ Montreal/ 
 Canada/ July 23-27/ 1973/ Donald A. Jardine/ ed./ North 
 Holland Publishing Company/ 1974 
 
 C33D Alice Ray/ "Habi t abi I i t y Information System/ Final Report/ 
 Pt . 1/" Champaign/ Illinois/ June 1975/ (Mimeographed) 
 
 [347 K. A. Robinson/ "DMS-110U/ An 
 Database/ Vol. 6/ No. 2/ pp. 8-14 
 
 ln-Depth Evaluation/" 
 
 [35] Gerard Salton/ Dynamic Information and Library TtQce ssina/ 
 Prentice -Hall/ Inc./ Englewood Cliffs/ New Jersey/ 1975 
 
 [363 Gerard Salton/ ed./ The SMART. Retrieval Sy_stem/ Prentice- 
 Hall/ Inc./ Ennlewood Cliffs/ New Jersey/ 1971 
 
 [37] G. Michael Schneider and Edouard J. Desautels/ "Creation of 
 
 a File Translation Language for Networks/" Information 
 
 Systems/ Vol. 1/ No. 1/ Penamon Press/ Great Britain/ 1975/ 
 pp. 23-31 
 
 [38] Michael E. Senko/ "Information Systems: Records/ Relations/ 
 Sets/ Entities/ and Things/" Information Systems/ vol. 1/ 
 No. 1/ Pergamon Press/ Great Britain/ 1 9 ? 5 / pp. 3-13 
 
 [39] William Howard Stellhorn/ "An Experimental Information 
 
 Retrieval System/" Department of Computer Science Report No. 
 
 657/ University of Illinois at Urbana- Champa i gn/ Urbana/ 
 Illinois/ July 1974 
 
 [40] Michael Stonebraker/ 
 Tutorial/" 1975 
 
 Getting Started in INGRES/ 
 
52 
 
 CERL HDB Recommendations 
 
 [413 Michael Stonebraker, Eugene Wong, Peter Kreps and Gerald 
 Held, "The Desiqn and Implementation of INGRES," Electronics 
 Research Laboratory Memorandum No. FRL-M577, University of 
 California, Berkeley, Berkeley, California, January 1976 
 
 [423 Roger K. Summit and Oscar Firschein, "Document Retrieval 
 Systems and Techniques," in Annual Review of Information 
 Science, Carlos A. Cuadra, ed., American ^Society for 
 Information Science, Washington, 1974, pp. 286 331 
 
 £433 Daniel J. Tanner, 
 
 Datamation, Vol. 21, 
 
 User Ratings of Software Packages, 
 No. 12, December 1975, pp. 133-154 
 
 [441 
 
 Massachusetts Institute of Technology, "Janus pinner's 
 Manual," Massachusetts Institute of Technology, Cambridge, 
 Massachusetts, 1975, (Draft) 
 
 [453 Massachusetts Institute of Technology, "Janus Us ^* s 
 Manual," Massachusetts Institute of Technology/ Cambridge, 
 Massachusetts, 1975, (Draft) 
 
 L463 
 
 P. Thorpe and M. Cocks, "TOTAL— Mar ket Leader in Independent 
 Database Management Systems," Database Journal, Vol. 6, No. 
 1, 19 76, pp. 2-8 
 
 [473 
 
 R L welsh, "User Manual for the Computer-Aided 
 Environmental Legislative Data System," U.S. Army 
 Construction Engineering Research Laboratory Interim Report 
 E-78, U.S. Army Construction Engineering Research 
 Laboratory, Champaign, Illinois, November 197* 
 
 [483 Martha E. Williams, "Use of Machine-Readable Data Bases, 
 
 in 
 
 Annual Review of Information Science, Carlos A. Cuadra, ed., 
 American Society for Information Science, Wasnington, 19/A, 
 
 PP 
 
 221-2^4