College and Research Libraries


RANDOLPH E. HOCK 

Providing Access to Externally Available 

Bibliographic Data Bases in an 

Academic Library 

The ready availability of externally processed bibliographic data 
bases has made it possible for an academic library to provide com-
puterized searches on a large number of data bases with a very small 
initial investment and utilizing its own personnel. The experience of 
the University of Pennsylvania Libraries has confirmed that such an 
approach is indeed feasible. This article discusses the approach, ques-
tions and problems encountered, and the factors considered in their 
resolution. Also discussed are the role of the data services librarian, 
the costs incurred, and some observations as to the philosophy of the 
approach, with particular attention to the integration of the service 
into the reference department. 

THE INCREASING AVAILABILITY of large 
machine-readable data bases has for sev-
eral years served as an indicator to li-
braries that a means of greater infor-
mation retrieval services to their users 
was forthcoming. The role of libraries 
in providing this type of service was dis-
cussed in a recent article by Richard 
DeGennaro.! In that article he wrote 
about the possibilities of the library's 
serving as a "broker" rather than as a 
processor for data bases. He pointed out 
that libraries will play a key role in pro-
viding access to data bases, but that the 
in-house processing approach is not fea-
sible due to the high cost involved and 
the nature of the demand. Also stressed 

At the time of the preparation of this 
article, Randolph E. Hock was data services 
librarian, University of Pennsylvania, Phil-
adelphia. He is now pursuing further grad-
uate study. 

208/ 

was the likely trend toward the utiliza-
tion of a mixture of batch-mode and 
on-line services, each serving specific 
functions. 

In the past two years the University 
of Pennsylvania Libraries, as well as 
some other academic libraries, have been 
pursuing this approach. The general 
goal has been to provide computerized 
literature searches for the university 
community, with a minimum invest-
ment, by taking advantage of currently 
available external services. 

In this article an attempt will be 
made to summarize the situation faced 
by academic libraries undertaking such 
an approach. The emphasis will be on 
the decisions which are necessary, the 
costs involved, and some general obser-
vations which should be of help to 
those contemplating accessing such sys-
tems. Technical details regarding the 
systems themselves will be avoided, since 



Access to Bibliographic Data Bases j 209 

such information is available elsewhere. 
An excellent up-to-date source of de-
tailed information about on-line sys-
tems is the recent book by Lancaster and 
Fayen. 2 

At the outset of projects of this type, 
certain policy decisions become neces-
sary which are potentially crucial to the 
success of the endeavor. Among the first 
that arises is the question of where 
within the library structure such a ser-
vice should be located. 

THE SERVICE AS A REFERENCE FUNCTION 

After consideration of the various al-
ternatives (organizationally and phys-
ically) within the library system, the 
conclusion has been reached that the 
reference department is the most rea-
sonable location. There are two primary 
reasons for this. First, providing biblio-
graphic services is a reference function; 
and, second, in order to give users fullest 
advantage of total library services this 
office should not be separate from, or 
in competition with, other reference 
functions of the library. To further 
this "integrated'' approach, it has been 
found feasible and desirable to involve 
the entire reference staff, particularly 
by delegating responsibilities for the 
searching among the staff, each special-
izing in one or more data bases. 

Alternatively, the service could be lo-
cated within the library's systems office, 
the rationale for this approach being 
the computerized origin of the services. 
However, for the reasons given above, 
the fact that handling these services re-
quires no systems experience, and the 
fact that the systems office is usually re-
mote from the public services division 
of the library, the former approach is 
preferred. 

TYPES OF SYsTEMS AvAILABLE 

The prospective "broker" of external-
ly processed data base services is con-
fronted with three main categories of 
service which can be used either sepa-

rately or in combination. These are 
batch-mode (off-line) services, on-line 
bibliographic services, and on-line non-
bibliographic (''statistical") services. 

The batch-mode services are the 
easiest (for the librarian) to use. Pro-
viding access to a large num her of these 
data bases can be as uncomplicated as 
collecting data base directories 3- 6 and 
then providing this information to users 
as the occasion arises. 

The on-line data bases require more 
effort to access, considerably more par-
ticipation on the part of the librarians, 
and a number of other considerations 
which will be discussed below. However, 
the actual amount of service which can 
be provided to the user will not only be 
greater, but is more likely to be utilized. 
The second type of on-line service, sta-
tistical data bases (such as Predicasts, 
the National Bureau of Economic Re-
search Time Series Data Bank, and 
STOCKRETRIEVE), presents more 
complicated problems, not in terms of 
the actual access, but in the on-line ma-
nipulation required due to the contents 
of the data base. The advantages are 
that the material retrieved is the actual 
information the user is looking for, 
rather than bibliographic references. 
This specificity of information has an 
additional drawback, however, in that 
each data base is likely to be of interest 
to but a small portion of the user com-
munity. 

uTILIZING BATCH SERVICES 

Following considerations of the above 
points, the next step is the actual im-
plementation of services. The discussion 
which follows will attempt to generalize 
where possible, but will also rely on pre-
sentation of actual situations encoun-
tered and experiences gained. As indi-
cated, taking advantage of batch ser-
vices involves, first, the collection of in-
formation on those available. In addi-
tion, the library may also become in-
volved in promoting the services and as-



210 j College & Research Libraries • May 1975 

sisting users in choosing the appropriate 
data base, making the contacts, filling 
out the request forms, and evaluating 
results. 

Unfortunately, none of the data base 
directories, though helpful, contain all 
the information which is frequently 
called for in fulfilling the above func-
tions. They do, nevertheless, serve as an 
adequate source for obtaining further 
information and price lists. 

Following this approach, information 
was compiled on those data bases which 
were readily available, and which were 
likely to be of interest to a number of 
potential users (by <<available" it is 
meant that they could be accessed in al-
ready processed form simply by contact-
ing a center either by mail or by 
phone). No attempt was made to pro-
duce a total list of either all data bases 
or all processing centers. It was found 
that twenty-nine available data bases 
might be of use and that these were 
available either directly from the data 
base producers or from one or more of 
six processing centers. 

This information was gathered, and 
considerable time was spent in the pro-
motion of the services. Working at this 
point only with the batch services, the 
promotional emphasis had to be placed 
on selective dissemination of informa-
tion ( SDI) services rather than the 
retrospective searches since the prices 
for SDI ($40/year and up) were much 
more appealing than the $200 which was 
typical for a four-year retrospective 
search. After a few months, only a 
handful of people had subscribed to 
such searches. The reasons given for not 
subscribing by most people contacted 
were that funds were not available and 
that they were satisfied with their pres-
ent information gathering methods. An-
other reason, which seemed obvious, 
though not stated directly, was that SDI 
services have yet to prove their value to 
university people. 

SELECI'ION OF ON-LINE SYSTEMS 

Besides the limited demand for those 
services which can be provided in batch 
mode, several factors indicate the ad-
visability of utilizing on-line services 
in addition to the batch services, in 
spite of the greater degree of involve-
ment demanded of the library. These fac-
tors are: ( 1) the cost of searches: on-
line retrospective searches (in most 
cases) are significantly less expensive in 
comparison to the batch searches; ( 2) 
the turnaround time: a few days or im-
mediately for on-line, a few weeks for 
batch; ( 3) the interactive capability: 
only available on-line; and ( 4) the 
rapid increase in the number of on-line 
data bases. 

More specifically in regard to the 
price differential, to get a typical retro-
spective search done on Chemical Ab-
stracts for 1970-1973 would cost $640 
from one leading center and $400 from 
another. The same search could be done 
for from $5 to $40 through the on-line 
services. A typical retrospective search 
for the NTIS files would cost $50 from 
NTIS directly, $65 from one center, and 
$50 from another, but less than $15 
through the on-line services. (Figures 
for the batch services are based on the 
promotional literature supplied by the 
centers. Figures for on-line costs are 
based on searches actually done. ) 

Once the choice has been made to im-
plement on-line services, the next deci-
sion necessary is which system ( s ) to ac-
cess and which of their data bases to 
use. ' At the present time there are three 
on-line bibliographic literature search-
ing systems which are actively promot-
ing their services. These are Lockheed 
Information Systems, New York Times 
Information Bank, and Systems Devel-
opment Corporation ( SDC). 
· Lockheed and SDC stand out imme-

diately as likely choices because of the 
data bases they offer, the range of sub-



Access to Bibliographic Data Bases / 211 

ject areas they cover, and the lack of 
any subscription charge or minimum 
fee. Based on the areas of study of the 
graduate students currently enrolled at 
this university, it was found that one or 
more of the data bases offered by those 
two systems could be of potential use 
to approximately 48 percent of that 
group. 

Likewise, the potential user group for 
the New York Times Information Bank 
is quite large, and this data base is con-
sequently quite appealing. The major 
difference, though, in accessing this sys-
tem as compared to the other two is the 
minimum investment required. Regard-
less of how little the system may be 
used, a minimum monthly cost of some-
thing in excess of $300 is incurred. 

Due to this necessary financial com-
mitment for the New York Times, utili-
zation of that service by this library was 
delayed until some experience was 
gained with the other two systems. How-
ever, within a few months after the im-
plementation of the first two systems, 
the New York Times was also accessed. 
The decision was precipitated by the an-
nouncement of a considerable decrease 
in minimum charges. 

PRICING STRUCTURE 

The primary factor to consider when 
deciding what to charge users for 
searches is how much the library is will-
ing to spend of its own money. If the 
amount is equal to what would be spent 
in developing one's own information 
retrieval system, the library can prob.a-
bly give away hundreds of searches per 
year absolutely free. Disregarding that 
option, the library can decide to offer 
free searches and limit demand in any 
one of several ways. This approach 
serves the very useful function of stim-
ulating use of the service. However, re-
sponse to anything free is usually not 
a good indicator of either potential de-
mand or actual need. Getting some feel 

for these two factors is of considerable 
importance at this stage in the develop-
ment of these services. 

At the opposite extreme, a library can 
decide to recover all costs including con-
nect time, communications, personnel, 
space, etc. This approach has the disad-
vantage of putting the cost of a search 
out of the reach of the graduate stu-
dents who perhaps need the service more 
than any other group. Also, if total cost 
recovery is a valid rationale for this ser-
vice, is it also a valid rationale for all 
library functions? 

Taking the approach that the searches 
should be partially subsidized, a reason-
able pricing structure, which has proved 
workable, is, first, to offer the services at 
a flat rate for a few months, and then, 
at the end of that period, when a nu-
cleus of subscribers has been created, to 
change the pricing structure to reflect 
costs more accurately. In both stages a 
criterion should be that the price sched-
ule should be easy for the potential user 
to understand. This approach was tried, 
using $10 per search as the flat rate (a 
''search" being defined as coverage of 
one data base with output limited to 50 
citations with abstracts, or 100 citations 
without abstracts). Additional output 
was offered at $.10 and $.05 respectively 
per citation. This price structure cov-
ered the direct costs ( connect time and 
communications) of some searches and 
fell far short for other searches. How-
ever, it allowed the expenditure of the 
approximate amount of "seed money" 
anticipated in order to advertise the ser-
vice. 

This price schedule was maintained 
through one term, at which time the 
structure was changed to give the user 
two options, designated as "standard 
searches," and "special searches." With 
the standard search, the user is charged 
a flat rate and is allowed to choose up 
to a certain number of descriptors (the 
rate and number of descriptors varying 



212/ College & Research Libraries • May 1975 

with the data base, but typically being 
$10 and ten descriptors). Output of up 
to fifty citations is included in the base 
price. For these searches the user is not 
present. For the special searches the 
user is present and is charged the actual 
cost of connect time and printing. As 
with the standard searches, the search 
strategy and input terms are worked out 
ahead of time, but are modified to 
whatever extent the user desires as the 
search progresses. 

PROMOTION AND PROCEDURES 

Since the type of service being offered 
is one with which members of the aca-
demic community are generally not fa-
miliar, but one for which the need 
should at least be explored, it is felt 
that a fairly vigorous promotional pro-
gram should be undertaken. This pro-
gram can involve the campus news me-
dia, flyers, posters, displays, direct per-
sonal contact, contact through depart-
mental librarians, and word-of-mouth 
referrals among users. Expenses for all 
of these approaches can easily be limit-
ed to personnel time and the library's 
own duplicating services. 

One other approach which is being 
tried is to make the grant applicants 
aware of the services at the time they 
are preparing their budgets. To accom-
plish this, an appendix describing the 
service was prepared for the handbook 
distributed by the university office which 
handles grants. The appendix describes 
the retrospective searches, the SDI 
searches, and also the "Research in 
Progress" searches which are available 
through the Smithsonian Science Infor-
mation Exchange. 

In developing procedures, the ques-
tion as to v;rhat extent to utilize user par-
ticipation has been one of the most dif-
ficult determinations to make. The ex-
tremes would have been either a totally 
delegated search, in which the requestor 
is not present, or a totally nondelegated 
search, in which the intermediate (the 

library staff member) need not be pres-
ent. The obvious disadvantage to the 
delegated search is that both recall and 
relevance can suffer. The advantage is 
that the search can be done in less time 
(with resulting decrease in cost). The 
disadvantages to the nondelegated 
search are the necessity for training ev-
ery requestor in use of the service (at 
great cost), and the likelihood that 
many, if not most, occasional users 
would not be able to take full advan-
tage of the systems capabilities (with 
a resulting decrease in recall). The ad-
vantage is an ultimately better perform-
ance for the frequent user (if any in-
dividual is willing to spend the money 
and time to become really familiar with 
the system). One solution is to take an 
intermediate approach and have the ref-
erence staff serve as "middlepersons," 
but to try to have the user present. This 
would result in lowest training costs, the 
availability of an experienced searcher 
for every search, and the availability of 
the requestor to give feedback as the 
search progresses. The impression has 
been gained that requestors who were 
actually present for the search seem fre-
quently more satisfied than those who 
opted to delegate the search. This de-
gree of satisfaction seems to be due not 
only to the increased relevance of the 
retrieved set but also to the fact that the 
requestor has a better understanding of 
the capabilities and limitations of the 
system. The usefulness of the "middle-
person" approach has also been indicat-
ed by Borman and Mittman. 7 

The policy of encouraging the re-
questor to be present for each search 
has been found to have several short-
comings: ( 1 ) some users find it incon-
venient to be present; ( 2) the total 
turnaround time is usually greater due 
to the necessity of waiting for a time 
convenient for both the searcher and 
the requestor; ( 3) some searches are so 
simple that the requestor's presence is 
unnecessary; and ( 4) connect times are 



Access to Bibliographic Data Bases / 213 

usually much greater with the requestor 
present, with the consequent higher cost 
of searching. It is recognized, however, 
that for many searches direct user par-
ticipation is still essential for most effi-
cient retrieval. This conflict can be re-
solved by offering the two options de-
scribed in the section on pricing. The 
standard search (flat rate, user not pres-
ent) permits availability of quick, inex-
pensive searches, especially for students, 
and the special searches permit more 
thorough searching utilizing more fully 
the total system capabilities. 

Another procedural decision is how 
many searchers to use. In most cases the 
volume of searches will be small enough 
to require only one or two searchers, but 
on the other hand, there are two main 
advantages to distributing the responsi-
bility among most or all of the refer-
ence staff. First, it helps to keep the 
service more integrally connected with 
the reference department; and, second, 
variation in content and indexing of 
the data bases is so great that the more 
data bases an individual attempts to 
handle, the less intimately each can be 
known, and the less efficient the retrieval 
process. This approach has been met 
with enthusiasm by reference staff mem-
bers, even though most have had no 
prior experience with computers. 

FUNCITONS OF THE 

DATA SERVICES LmRARIAN 

The function of this person (called 
information services librarian in some 
institutions) can be broken down into 
five areas: coordination, promotion, 
training, searching, and consultation. 

Coordination includes development 
of procedures, liaison with and among 
searchers, and liaison with departmental 
librarians and other related services on 
campus. 

Promotion is the one function which 
is probably most alien to many librari-
ans. Nevertheless, if it is indeed felt 
that the services being offered are time 

saving and worthwhile, one should ei-
ther be willing to put considerable effort 
into their promotion, or rethink the 
premise. Over the next few decades, pat-
terns of scientific communication are 
sure to change in this direction, but if 
these services are not exploited at pres-
ent, many valuable resources are going 
to be wasted in the interim. 

Training in the use of the on-line 
systems consists essentially of digesting 
the manuals and accumulating experi-
ence. Once one person within an institu-
tion has become familiar with the sys-
tems, that person can train others easily. 
Training courses are offered by both 
Lockheed and SDC, and experience has 
shown these to be helpful; but if one 
is capable of reading the manuals 
(which assume only a rudimentary 
knowledge of systems terminology), one 
can proceed from there and, within a 
couple of hours of connect time, be 
well on the way to proficiency. Again, 
the knowledge of the operation of the 
systems is but one part of what is to be 
learned, the other main part being 
knowledge of the indexing and contents 
of the individual data bases. Manuals 
for both SDC and Lockheed could be 
of greater help if they included details 
about the indexing policies followed 
(or allegedly followed) by the data 
base producers, and also some details 
about the posting procedures followed 
in putting those data bases onto the on-
line indexes. 

It was found that an average of 3.3 
hours of connect time was necessary per 
searcher in preparation for his or her 
first actual search. Additional training 
time was accumulated in the first few 
searches and is reflected in the differ-
ence in average search time for the first 
month and the second. The average 
time decreased by eleven minutes dur-
ing that period. Charging this difference 
to training costs gives a total training 
( connect) time of less than four hours 
per person, or $140--$280 of connect 



214 /College & Research Libraries • May 1975 

time (depending on the data bases 
used). 

The consultation function served by 
the data services librarian tends to serve 
simultaneously as an in-depth reference 
function. Primarily, the consultation is 
to provide potential users with informa-
tion about the computerized searches, 
but this information is more valuable 
to the user if supplemented by further 
suggestions as to the approach to his or 
her literature problem. Such personal 
attention can be had at the reference 
desk, but is frequently qualitatively and 
quantitatively insufficient due to lack of 
time. 

Providing the information about the 
computerized searches requires that the 
data services librarian be familiar with 
what services are available, what the 
output looks like, the turnaround times, 
which data elements are included, the 
types of literature covered, and specif-
ically which journals are covered. For 
the batch services this information can 
be gotten mainly from the promotional 
literature of the centers. For the on-
line services, this information is avail-
able from the manuals, the promotional 
literature, and the hardcopy of the data 
bases. 

COSTS 

A cause for considerable apprehen-
sion when approaching these services is 
how to anticipate real costs correctly. Al-
though the systems sound relatively in-
expensive, few figures are available ex-
cept from the suppliers. Fortunately, 
there are no large cost categories which 
are not readily anticipated. Costs can be 
controlled to any degree a particular li-
brary desires, depending upon how 
quickly the library wishes to offer the 
services, how many of the data bases are 
offered, how many searchers are trained, 
the sophistication of the terminal used, 
and the degree to which the library is 
willing to subsidize the services. 

Fixed costs include equipment rental 

(or purchase), service contracts for 
equipment, training time, and overhead. 
Equipment can be rented for about $80 
per month on up, and service contracts 
are about $20 per month. Overhead is 
the same as for other library operations. 
Personnel costs will vary, depending 
upon how quickly and how deeply the 
library wishes to become involved. 

The variable costs depend upon how 
many searches are done and how long 
the searches take. For a typical search, 
about thirty minutes of the coordina-
tor's time is required (talking with the 
requestor, handling printout, billing, 
etc.). A search typically requires from 
ten to thirty minutes for its execution 
(including discussion, logging on, and 
connect time). Printing costs vary from 
$.05 per page up to $.25 per citation. 
Typical searches retrieve between ten 
and sixty citations. It has been found 
that the average direct cost-per-search 
on the ERIC data base, which is one of 
the least expensive, but also the most 
popular, is less than $12. Searches on the 
other data bases are greater, approxi-
mately in proportion to the cost of con-
nect time. 

SUMMARY 

Experience with accessing externally 
available bibliographic literature search-
ing services indicates that such an ap-
proach is feasible (perhaps even obliga-
tory) in the academic library, without 
waiting for further developments or ex-
tensive studies. Costs are reasonable, and 
perhaps of more importance, subject to 
whatever controls may be applied. User 
acceptance is good, though not over-
whelming, and the approach of supply-
ing ''seed money" seems to be effective 
since an increasing proportion of users 
have been referred to the service by for-
mer users. 

As to how these services fit in with 
users' needs, it is felt that many under-
graduates can use the services as a basic 
source for some of their papers. For 



Access to Bibliographic Data Bases / 215 

graduate students, faculty, and other re-
searchers, the service can fulfill an im-
portant function as a supplement to 
their usual information retrieval meth-
ods ( as indicated by Back8 ) • 

As more experience is gained, as more 
formal studies are completed, and as 
suppliers become more aware of the 

academic market, the ease and efficiency 
with which the services can be provided 
will increase. The present state of the 
art is such, however, that any academic 
library with sufficient interest can suc-
cessfully serve as a "broker" for these 
services. 

REFERENCES 

1. Richard DeGennaro, "Providing Biblio-
graphic Services from Machine-Readable 
Data Bases-The Library's Role," Journal 
of Library Automation 6:215-22 (Dec. 
1973). 

2. F. W. Lancaster and E. G. Fayen, Informa-
tion Retrieval On-Line (Los Angeles: Mel-
ville, 1973). 

3. ASIDIC Survey of Information Center Ser-
vices. 

4. Organization for Economic Cooperation and 
Development, Inventory of Major Informa-
tion Sy8tems and Services in Science and 
T echnology (Paris: OECD, 1971). 

5. John H. Schneider, Marvin Gechman, and 
Stephen E. Furth, Survey of CommerciaUy 
Available Computer-Readable Bibliographic 

Data Bases (Washington, D.C.: American 
Society for Information Science, 1973). 

6. Vivian S. Sessions, ed., Directory of Data 
Bases in the Social and Behavioral Sciences 
(New York: Science Associates/Internation-
al, 1974). 

7. Lorraine Borman and Benjamin Mittman, 
"Interactive Search of Bibliographic Data 
Bases in an Academic Environment," Jour-
nal of the American Society for Information 
Science 23:164-71 (May/June 1972). 

8. Harry B. Back, "What Information Dissemi-
nation Studies Imply Concerning the Design 
of On-Line Reference Retrieval Systems," 
Journal of the American Society for Infor-
mation Science 23:156-63 (May/June 1972).