Intelligent Technologies in Library and Information Service Applications (ASIS&T Monograph series)

Citation

Malinconico, M. (2003), "Intelligent Technologies in Library and Information Service Applications (ASIS&T Monograph series)", Program: electronic library and information systems, Vol. 37 No. 1, pp. 63-64. https://doi.org/10.1108/00330330310460635

Computers that could exhibit human‐like intelligence have long fascinated the general public and challenged computer scientists. With each new computer advance we always seem on the verge of achieving this goal. However, the goal has proven to be as elusive as a will‐o’‐the‐wisp. In part it may be because we have no ready definition of what human‐like intelligence is. Is it a system with which one can interact yet not know for certain whether or not one is interacting with a human or a machine – ELIZA? Is it a system that can understand and respond to spoken commands – many telephone menu systems? Is it a system that can apply logic and a set of rules to solve problems or to analyse situations – expert systems, or knowledge‐based systems? Is it a system that can visually identify objects – factory robots? Is it a system that is aware of context and can guess the intentions of its users and is able to offer unsolicited assistance – the annoying Microsoft paper‐clip? Is it a system with a will of its own – HAL from 2001? Or, more modestly, is it a system that is capable of accomplishing to the satisfaction of its user specific tasks by exercising independent, unprompted judgment? With the possible exception of the last example instances of such systems exist and are actually in use – and in some instances in widespread use.

Computers have unquestionably proven their ability to retrieve structured information from even very large databases. However, the problem becomes less wieldy when the information is unstructured or loosely structured and when there are many sources to search. Consequently, there is a continuing interest in computerised facilities, or agents, that can, with relatively little human guidance, search those sources and report their results. The rapid growth of information sources on the Internet has increased the potential utility of this resource, but it has simultaneously made exponentially more difficult the problem of locating the precise information one seeks – or of determining when an unsuccessful search should be concluded. Thus, librarians and information specialists have a strong vested interest in intelligent technologies that can facilitate the searching of the fecund, though largely untamed wilderness that is the Internet. In fact, relatively primitive intelligent agents roam the Web in a relatively successful attempt to index at least a fraction of the resources that are mounted on the Internet.

Librarians are understandably impatient for additional technological advances that will facilitate their ability to discover and use the resources that are available in cyberspace. Thus, a treatise on Intelligent Technologies in Library and Information Service Applications by two respected information scientists would seem certain to garner considerable attention among information specialists of all kinds. F.W. Lancaster and Amy Warner have written a book with that title. They have undertaken a wide‐ranging search of the databases that index the literature of library and information science. They cast their net very wide, e.g. they searched for materials on broad topics such as artificial intelligence, expert systems, neural networks, etc. They then used the result of these searches to formulate more refined searches of databases that might yield results of interest to the “information services community”. These refined search strategies were also applied to databases that index materials in science and engineering. The authors obtained copies of materials that seemed most promising as sources of information of developments that would be of use or interest to information specialists.

Systems have been developed or proposed in many areas of librarianship: cataloguing, subject indexing, acquisitions and collection development, reference services (including referral of users to appropriate sources, database selection, and assistance with the formulation of database search strategies). A number of systems have been attempted to aid indexers and to train indexers. However, Lancaster and Warner note that “the algorithms that extract or assign terms exhibit little that can be considered truly intelligent”. Systems have also been proposed or developed that automatically generate indexing terms, but these are far from serious realisation. The situation is even bleaker in the other areas mentioned. A sizeable number of systems are identified, but none seem to have progressed beyond the demonstration stage. In sum, Lancaster and Wagner conclude:

Many library‐related applications of expert systems or “intelligence technologies” have been discussed in the literature, but this is extremely misleading. Systems that have progressed to an “operational” state – i.e. are functioning on a daily basis and providing a real service to library staff or users – are almost non‐existent. The literature is full of article titles suggesting that systems exist when in fact they never did or did only in a very limited sense.

There are also a number of applications of intelligent technologies which, while not intended specifically for, are, nonetheless, of use or of interest to the library and information specialist community. These include intelligent text processing (e.g. information retrieval, machine translation, text categorisation, text extraction, text summarisation, text generation, etc.); intelligent agents (software that can retrieve information from multiple sources without human intervention); intelligent interfaces; and data mining (systems that analyse patterns in data and extract information) and knowledge discovery (systems that transform data into previously unknown or unsuspected relationships that may be used to make future projections or predictions). Here again Lancaster and Wagner found very few operational systems that exhibit truly intelligent – as opposed to algorithmic – behaviour.

Finally, the authors also surveyed intelligent systems developed for use in other fields, e.g. manufacturing, finance, law, education, health care, human resources, etc. Some interesting systems with potential applications to library and information service applications are identified, the most notable of which are Help Desk systems. These systems often employ rules databases and inference engines. Several are commercially available products. Of particular interest is that systems are appearing that can be utilised by users themselves, without the need for intermediaries.

The results of the study do not seem very encouraging. With only very rare exceptions Lancaster and Warner found that the applications of intelligent technologies described in the published literature were either speculation, designs for systems or prototypes, many of which had already been abandoned. However, the evidence offered too often consists of citations to work published as long as ten years ago, for example:

Kulich (1992) claims that voice‐recognition technology is not sufficiently advanced to allow for useful speech‐to‐text translation in the short term (p. 97).

Hall (1992) discusses work undertaken to develop an expert system for selection of periodicals, but there is no evidence that it was actually implemented (p. 26).

The recognition of handwriting continues to be of research interest (see, for example, Nakagawa et al., 1993) … In limited handwriting recognition tasks, such as the recognition of ZIP‐code digits, a 96 per cent success rate is claimed (see, for example, Fontaine and Shastri, 1993), but general‐purpose systems for recognizing handwritten characters are not on the immediate horizon (p. 104). [This may come as a surprise to the manufacturers of hand‐held PDAs, etc.]

Rudnicky et al. (1994) surveyed the state‐of‐the‐art in this field [speech technologies] …

Sparck Jones (1991) is one writer who has stated the case most clearly … [S]he maintains that we should not overestimate the potential contribution of artificial intelligence (p. 126).

The conclusions that Lancaster and Warner reach are probably valid; however, the evidence they offer is not immediately persuasive. Admittedly, research that demonstrates a negative conclusion – “There are few, if any, viable applications of intelligent technologies for library and information service applications at the present time” – is as useful and valid as research that is evidence for a positive conclusion. However, this could have been accomplished in the form of a well‐reasoned – and surely much briefer – journal article. The lengthy descriptions of the many unsuccessful, vintage systems and projects that are laboriously noted in this work do not seem to aid very much in this endeavour. Nor are they particularly valuable to someone seeking to learn about the present state‐of‐the‐art of the application of intelligent technologies in library and information service activities. In fact, Intelligent Technologies in Library and Information Service Applications resembles a dissertation that was completed a number of years ago and is now being published as a monograph.