157 LIBRARY NETWORK ANALYSIS AND PLANNING (LIB-NAT) Maryann DUGGAN: Director, Industrial Information Services Program, Southern Methodist University, Dallas, Texas A preliminary report on planning for networl< design undertaken by the Reference Round Table of the Texas Library Association and the State Advisory Council to Library Services and Construction Act Title III Texas Program. Necessary components of a network are discussed, and network transactions of eighteen Dallas area libraries analyzed using a methodology and quantitative measures developed fm· this project. To be a librarian in 1969 is to stand at the crossroads of change, with a real opportunity to put libraries and professional experience to work on immediate problems of today's world. In mobilizing total library resources for effective service to a variety of patron groups in a variety of ways, the librarian has at hand an exciting new tool of great potential and equally great challenge: the library network LIBRARY NETWORKS AND REFERENCE SERVICES Networks and all that they imply are simply an extension of good ref- erence services as they have been practiced for years, but their existence and potential capability require redefinition of the reference function, which, being no longer limited to one collection, has been given new dimensions of time, depth and breadth. Networks, and the inter-library cooperation they require, offer an op- portunity to combine materials, services and expertise in order to achieve more than any one library can do alone. In this case, the whole is greater than the sum of its parts, for each library can offer its particular patron group the total capability of the network, including outside resources not previously available. 158 Journal of Library Automation Vol. 2/ 3 September, 1969 With the new tool of library networks, it is possible to provide respon- sive, personalized, in-depth reference service, and to provide it so rapidly that a patron can receive a pertinent bibliography covering his desired topic within an hour of his original inquiry. The reference librarian be- comes an expert in resources and resource availability at the national level. His reference desk becomes a switching center, at which he receives and analyzes inquiries, decides the level of service required, identifies available sources or resources that match an inquiry, transmits the latter ( restruc- tured to be compatible with the network language), conducts a dialog with the source, receives the response and interprets it to the patron. This procedure is not markedly different from what has been done for years in any reference library, but with greater potential the process must be more formalized and structured. Networks do require new expertise and crystallizing the reference phi- losophy. Clarification is needed as to 1) types or levels of reference serv- ices, and unit operations in reference services; 2) the role of in-depth subject analysis of reference queries; 3) decisions on alternate choices of sources and of communications links; 4) structuring of large blocks of resources to permit fast access; and 5) the role of each library in the network and its responsibility to the network. APPROACH TO NETWORK DESIGN The Reference Round Table of the Texas Library Association and the State Advisory Council to Library Services and Construction Act Title III Texas Program have been struggling with the challenge of inter-library network design for the past two years. This paper is written to share with reference librarians some of their preliminary findings and to urge the involvement of reference librarians in planning and developing networks and network parameters. For identification the project herein described is referred to as Lib-NAT, for Library Network Analysis Theory. Although only the author can be blamed for any faults of this "theory," many persons have contributed to the development of it. The Reference Round Table of the Texas Library Association has provided the forum for exploring and developing ideas on inter-library cooperation. Title III of the Library Services and Construction Act has provided the legal and financial impetus enabling the field testing of some of those ideas. Texas Chapter, Special Libraries Association, has sparked and catalyzed ideas and clarified needs. The State Technical Services Act provided the vehicle for experimental development of new approaches to reference services. Southern Methodist University provided the haven and ivory tower from which these new approaches could be tried under the cloak of academic respectability. But, of greatest importance of all, individual librarians, with vision and desire to be of service and willingness to try new things, have been the driving force in helping to develop new concepts of library use and purpose in the Texas area. Lib-NAT 159 The basic philosophy back of Lib-NAT is simply that any person any- where in the State of Texas should have access to any material in any library anywhere in the State through a planned, orderly, effective system that will preserve the autonomy of each library while serving the needs of all the citizens of the State. Particular needs of special user groups (such as the blind or the accelerated student or the industrial researcher) should also be identified and provided for in a cooperative mode through local libraries throughout the State. NETWORK COMPONENTS In the process of developing Lib-NAT, twelve critical components were identified that are essential to orderly, planned development of the objec- tives stated above. As a minimum, such a network must have the following: 1) Organizational structure that provides for fiscal and legal responsi- bility, planning, and policy formulation. It must require commit- ment, operational agreement and common purpose. 2) Collaborative development of resources, including provision for co- operative acquisition of rare and research material and for strength- ening local resources for recurrently used material. The develop- ment of multi-media resources is essential. 3) Identification of nodes that provide for designation of role speciali- zation as well as for geographic configuration. 4) Identification of primary patron groups and provision for assign- ment of responsibility for library service to all citizens within the network. 5) Identification of levels of service that provide for basic needs of patron groups as well as special needs, and distribution of each service type among the nodes. There must be provision for "refer- ral" as well as "relay" and for "document" as well as "information" transfer. 6) Establishment of a bi-directional communication system that pro- vides "conversational mode" format and is designed to carry the desired message/document load at each level of operation. 7) Common standard message codes that provide for understanding among the nodes on the network. 8) A central bibliographic record that provides for location of needed items within the network. 9) Switching capability that provides for interfacing with other net- works and determines the optimum communication path within the network. 10) Selective criteria of network function, i. e., guidelines of what is to be placed on the network. 11 ) Evaluation criteria and procedures to provide feedback from users and operators and means for network evaluation and modification to meet specified operational utility. 160 Journal of Library Automation Vol. 2/ 3 September, 1969 12) Training programs to provide instruction to users and operators of the system, including instruction in policy and procedures. The foregoing components of the ideal inter-library network (one so designed that any citizen anywhere in the state can have access to the total library and information resources of the state through his local li- brary) may be considered the conceptual model, or the floor plan from which the network of the program can be constructed. Although these twelve components might be labeled "ideal," they are achievable and they are within reach of the present capability of all libraries today. They have also weathered the unrelenting critique of 288 reference librarians in the March 27, 1969, TLA Reference Round Table ("The 1969 Reference Round Table Pre-Conference Institute: An Overview," Texas Library Journal, Vol. 45 (Summer 1969), No. 2.). During that Reference Round Table the twelve components were tested in a simulated network, using 42 cases. In this behavioral model actual, current inter-library practices were observed during game-playing in the simulated network. The expe- rience verified that the components outlined above are essential to the development of planned, cooperative, inter-library systems. ANALYSIS OF NETWORK TRANSACTIONS As part of the LSCA Title III project, and to test the twelve compo- nents, exploration was instituted into the existing inter-library relations among eighteen libraries of all types in the Dallas area to see how cur- rent practices compared with the ideal conceptual model The essential minimum requirement of a library is document transfer, i. e., the ability to supply a known item on request; and on-going inter-library loan trans- actions are a valid indicator of emerging network patterns in the current environment. This microscopic study of 1967 individual library loans among eighteen libraries of different types has provided a wealth of insight into network developments. As a pilot model it has offered a means of observing and studying existing practices, identifying problems, and experimentally eval- uating the effect of changes in the system or environment. More must be known about on-going inter-library transactions for the design of improved networks. In the attempt to find out who was attempting to borrow what from whom and how successfully requests were filled, the following vari- ables were considered: 1) Type of library, both borrowing and lending, such as academic, public, special, or public school. 2) Type of message format, i. e., telephone, TWX, TELEX, letter, or interlibrary loan. 3) Type of item requested in the transaction, such as monograph, serial, map, document. 4 ) Geographic location of borrowing and lending library, i. e., local, area, state, regional, national or international. Lib-NAT 161 The complexity of even a small pilot model required the formulation of some rigor in the analysis and the development of analytical tools and symbolic models. Figure 1, for example, is a symbolic model that permits comparison of two variables simultaneously, e. g., the type of library par- ticipating in the transactions and the geographic level of the participants. For modeling purposes, it was assumed all libraries fall into one of four 1 = Local 3 = State 2 = Area 4 = Re ion SWITCHING CENTERS Fig. 1. Symbolic Model of Inter-Library Networks. classes represented by the quadrants in Figure 1. Also it was assumed that each library can be identified as to a specific geographic level, as indicated by the numbers 1 through 6. In the analysis of the pilot model data it was observed that transactions occur among libraries of the same type and at the same geographic level, and between libraries of different types at different geographic levels. Figure 1 provides a symbolic model for conceptualizing these various types of transactions. Switching centers, represented on Figure 1 by the circles around the geographic numbers, participate in transactions at varying geographic levels, as well as between and among various types of library sectors. The role and the location of switching centers is an important aspect of Lib-NAT. 162 Journal of Library Automation Vol. 2/ 3 September, 1969 Within the framework of the symbolic model, the simple form of inter- library loan may be represented as a two-body transaction between the borrowing library and the lending library, as shown in Figure 2. Applying these transactions on the symbolic model of Figure 1 and considering both A B Fig. 2. Two-Body Transaction. type of library and geographic level, four general classes of two-body transactions can be identified: 1 ) Homogeneous vertical, i. e., between two libraries of the same type but at different geographic levels (Pt _..,.. P~; St _..,.. Sa) ; 2) Heterogeneous horizontal, i. e., between two different types of libraries at different levels ( Pt _..,.. A1; St _..,.. P1); 3) Heterogeneous vertical, i. e., between two different types of libaries at different levels (Pt _ ..,.. A4; sl _..,.. PG); 4) Homogeneous horizontal, i.e., between two libraries of the same type and the same geographic level (Pt _..,.. Pt; S2 - .... S2). The formulas serve as a shorthand symbolic representations of some typi- cal transactions of these four classes. The final report on Lib-NAT will contain statistical data on distribution of pilot model transactions by type and by geographic level, showing type interdependency and geographic dependency or self-sufficiency. Further analysis of the pilot model data revealed another type of trans- action, the three-body transaction, in which a third agent becomes in- volved. The third agent may act as a referral center, as illustrated in Figure 3, or as a relay center, as illustrated in Figure 4 ( SW indicates switching center) . Part of the Lib-NAT theory specifies that there is a dis- tinction between referral and relay, and that the latter is a valid function of a true switching center. Figure 5 illustrates the various types of possible three-body transactions with different geographic levels of switching among the different types of libraries. Which of these transactions is the most efficient or has the greatest utility is one of the basic design param- eters needing further analysis. It should be noted that the variable, of message format, that is, the channel of communication or type of com- munication link, has not yet been investigated in the symbolic modeling of these transactions. Lib-NAT 163 ..... .... A ... B c ~ .4~ t Fig. 3. Three-Body Transaction: Referral. ... .... A sw B Fig. 4. Three-Body Transaction: Relay. At • SWt ., St At • SWs ~ A4 Ps • SWs • Pt sc1 ~sw2 .... sc2 p2 .. P1 ~sw3 .. p3 p2 ... P1 ~sw3 ~ A3 p2 • P1 ~sw3 .. SW1 ~s1 Fig. 5. Three-Body Transactions at Various Geographic Levels. - .. . . 'I 164 Journal of Library Automation Vol. 2/ 3 September, 1969 NETWORK CONFIGURATION Another very important design parameter is the network configuration or organizational hierarchy specifying the communication channels and message flow pattern. Figure 6 illustrates symbolically a non-directed con- figuration of communication. If each dot represents a node in the network ( i. e., a participating library), and each line represents a communication link, it can be seen that each node can communicate directly with every other node, providing (or requiring) a total of fifteen links among the six nodes . N·l C = N (-2-) =15 Fig. 6. Non-Directed Network . By contrast, Figure 7 illustrates a directed configuration to which the six nodes are interconnected through a switching center and requiring only six channel links. In like manner, if a non-directed network desires to inter- face with a specialized center, such as the Library of Congress or a spe- cial bibliographic center or search center, a total of twenty-one channels is required (Figure 8), whereas a di- rected network can interface with a specialized center via only seven channels, as illustrated in Figure 9. __ .... ------ ------ jc~N-t=sl Fig. 7. Directed Network. Lib-NAT 165 Fig. 8. Non-Directed Network Including Specialized Center. Fig. 9. Directed Network Including Specialized Center. - 166 Journal of Library Automation Vol. 2/ 3 September, 1969 As local or area networks begin to develop, there will be a need for tying together two area networks to develop larger units of service. The interfacing of an original network of six libraries in one area with an ad- joining area network of SL"( libraries will result in the network configuration shown in Figure 10 in the case of a non-directed network, and sixty-six communication links among twelve nodes will be required. Whereas, if two directed networks of six libraries each desire to interface, a type of linkage requiring only thirteen channels may be envisioned (Figure 11). Which is the best type of network configuration? What are the decision parameters that should be considered in designing or planning network configuration? How can alternate configurations be evaluated ? Alternate channel requirements? And alternate geographic levels of switching? In the pilot model study, a mathematical model has been devised which can be used for simulating various configurations and channel capacities, Fig. 10. Interface of Two Non-Directed Networks. Fig. 11 . Interface of Two Directed Networks. Lib-NAT 167 thereby permitting some desired criteria function of network performance to be maximized or optimized. The details of the mathematical model will be published as part of the final report on Lib-NAT; in the meantime it can be said that this is a fascinating area of network analysis which will be useful to any group of libraries planning network configurations. The mathematical model-a multi-commodity, multi-channel, capacitated network model, developed by Dr. Richard Nance at Southern Methodist University as part of the Title III project-promises to have a high poten- tial application in network design and performance evaluation. It does require that the librarian make some hard-nosed decisions on operational and performance parameters of the inter-library systems discussed in the preceding article, but this is part of the challenge of Lib-NAT. MEASURES OF PARTICIPATION It is obvious that types of libraries, geographic level, types of transac- tions, various network configurations, alternate communication links and switching levels are all important in planning inter-library systems. Next it is necessary to take an in-depth look at the relationship between the individual participating library and the total network. In the pilot model study of eighteen libraries a noticeable difference appeared in the magni- tude and type of participation. In surveying only the two-body transac- tions, it was observed that some libraries were primarily borrowers and others primarily lenders, and some were heavy and some light. In pursuit of a quantitative method of representing these relationships some formulae were evolved which are helpful in understanding node/network dynamics. Starting with the individual library or node, let B .. equal the number of borrowing transactions originating at that node and L,. equal the num- ber of lending transactions; then L.. plus B,. will equal the total number of all transactions at that particular node. In like manner, looking at the total network (in this case all eighteen participating libraries), let Bt equal the total number of borrowing trans- actions originating in the network and L, the total number of lending transactions; then B, plus Lt will equal the total number of both types of transactions in the network. In the analysis of node/network dynamics, it was felt there should be some way of quantitatively expressing the individual node's dependency on the total network and also a way of expressing the relative degree of activity of each node. In other words, a participating library that was a net borrower (compared to its lending) was obviously more dependent on the network than would be a library that borrowed very little compared to its lending. The extent of dependency can be expressed as a node dependency coefficient calculated as follows: B. B,. + L,. Relative amount of borrowing compared to total node transactions .. .. 168 Journal of Library Automation Vol. 2/3 September, 1969 Among its other uses, the dependency coefficient of a node may give some insight into the extent to which it should share in network expenses, but the dependency coefficient alone should not be a final criterion, since magnitude of activity is of equal importance. For developing a method of quantitatively expressing activity of a node compared to total activity of the network a factor called the node activity coefficient may be calculated as follows: Relative activity of both types at _ one node compared to total activity Be + Lt in total network Bn + Ln Then, to quantitatively express the dependency of a given node on the network, one can calculate the node/ network dependency coefficient as follows: B B+L Fig. 12. 0. 0.6 0.5 0.4 0.3 o' o I 0.2 cp I 0.1 I I I 100 Be + Le > o. 5 = Net Borrower < 0.5 = Net Lender I I I I Q I I I I I ·I I I I I 200 300 400 B+ L Node Dependency Coefficient. I I I I I I B>L1 I B