Intelligent Technologies in Library and Information Service Applications

This is a short but valuable work, essential reading for all interested in, hopeful for, or sceptical of the application of artificial intelligence and expert systems in the library and information field. (The actual text, including appendices, is only 149 pages long.) Chapter 5 (“Conclusions and implications”) is almost worth the price alone. It is a masterly, concise account of the present state of and future hopes for AI and ES. It is abundantly clear that while there have been many proposals for AI/ES, and not a few experimental systems, there are very few actual working applications. Enthusiasts praise them but practitioners do not generally use them.

Intelligent systems ought, many of us feel, both to make the information worker’s life easier and to do things we cannot do, especially in the areas of indexing and retrieval. The authors point out that AI is interpreted in various ways: “behaving intelligently”, which few systems do, has to be distinguished from “[doing] things that human beings need intelligence to do”, where the prospects are brighter (and similarly with expert systems). Alas, even if the looser definitions are used the costs of implementation may, and usually do, make application totally uneconomic: the main expense lies in the human costs of knowledge acquisition and reducing it to system form. Moreover, “few librarians have the knowledge or experience needed to implement an expert system or to attempt to apply artificial intelligence techniques”; and “library applications are in many ways more complex than seemingly similar applications elsewhere”. Optimism in the 1960s and 1970s gave way to disillusionment in the 1980s. A third phase now seems to be on its way: a return to enthusiasm as a consequence of the enormous growth of the Internet and the Web, which “creates new applications, modifies existing ones, and facilitates collaborative projects in the implementation of systems”.

The authors go on to look at applications of advanced technologies in the library of the future. They believe that “the library cannot survive as a mere switching center, even a value‐added switching center”, and that it must act as an information filter, “[organizing] the universe of resources in such a way that those most likely to be of value to the user community are made most accessible”. A model that might achieve this is proposed. They recognise that “information processing tools of ever‐increasing sophistication will continue to be developed, and … some of these will be of direct relevance to librarians and other providers of information services”, but there is a danger of too great reliance being placed on intelligent systems, with a subsequent loss of uniquely human expertise. In short, watch developments, though with a highly critical eye – but do not try to develop systems on your own and do not believe half the enthusiasts say.

The four preceding chapters provide the evidence for Chapter 5, and review the existing scene. They are based on a truly exhaustive literature search – the 50‐page bibliography contains only a relatively small percentage of the items looked at – and replies to letters sent to relevant companies and authors. They deal respectively with “Applications within libraries and other information service operations”, “Applications closely related to library problems” (including intelligent text processing, intelligent interfaces and data mining), “Applications from other fields” (including help desks and medical diagnosis), and “General technologies” (speech technology, computer vision). All are highly informative.

There are one or two oddities. Archaeology nowadays hardly belongs with the humanities (p. 83). On pp. 48‐9 an “interesting hypermedia system” is illustrated by the question “Can you show and describe to me a fresco by Ambrogio Lorenzetti in Siena?”; the response generated is described as a “rather complete answer”, although it merely lists one such fresco without description and provides some additional irrelevant information.

One of the five appendices, “Criteria for deciding when an expert system approach can be justified”, is of particular interest, consisting as it does of two reprinted pieces, by Dreyfus and Dreyfus and by Quantrille and Lu respectively.

It is impossible to say what share each of the authors has had in writing this book: no seams are visible. Whatever contribution Amy Warner has made (probably most notably in the analysis of the literature), the work bears the hallmark of Lancaster: crystal clear in expression, highly readable and unafraid to express firm (and firmly founded) opinions. He has made a huge contribution to the library and information world over the years, not least by his influence on generations of students and co‐authors. This book is yet another notable addition to his oeuvre. Long may he continue to inform, irritate and inspire us.