expert system

An expert system is a computer system designed to capture the knowledge and behaviour of an expert in some problem domain; they are also known as knowledge-based systems for this reason. Typical domains are medical diagnosis or electronic repair, where the expert identifies problems and solutions to them from a limited set of data, making inferences from the combination of their knowledge of the domain and the data available. In such cases, the aim of the expert system is to reproduce this ability in a computer system, allowing people with less experience access to the expert’s knowledge. A typical expert system is composed of an inference engine and a knowledge base. The knowledge base consists of the rules and facts which compose the knowledge acquired from the expert, and the inference engine uses these rules and facts along with data entered by the user to draw conclusions.

Knowledge-based systems in libraries

Automation can do more than help librarians provide information to users. Sometimes computer applications might even replace the need for reference staff to serve as intermediaries between users and information. The development of knowledge-based systems or expert systems has this promise.

Knowledge-based systems is often abbreviated as KBSs. These systems have come out of the field of artificial intelligence. Library applications soon followed the initial work on artificial intelligence in the late 1950s.

A pioneer in the field of AI from Stanford University named E.A. Feigenbaum defines knowledge based, or expert systems as “an intelligent computer system that uses knowledge and inference procedures to solve problems that are difficult enough to require significant human expertise for their solution.” For humans to exhibit expert performance in any number of tasks they need extensive knowledge and experience. A KBS tries to emulate this performance by acquiring and using the same knowledge that the human expert has.

There are two main reasons that librarians are interested in developing KBSs. The first is to make their expertise available to customers at all times. The second is to save staff time in answering “routine” questions. This is important in times of decreasing resources. A system was developed as early as 1967 that was designed to retrieve the titles of works that could best answer biology reference questions. Several computer programs called expert systems had appeared by the mid-1970s. These early systems had limited flexibility, speed, ease of development and alternation, and portability. Some of these limitations were overcome by expert systems that could be mounted on microcomputers, and became available to customers in particular libraries.

A knowledge-based system consists of three things. First is a knowledge base of facts and heuristics related to the problem. Second is a control structure, or an inference procedure. This is to utilize the database in solving the problem. Third is a working memory, or database to keep track of the status of the problem.

Communication is required between computer programmers and experts in the field in order to develop expert systems. Computer programmers in this field are called knowledge engineers. Experts tell programmers how problems are solved and the heuristics that are used. The knowledge engineer takes the information and programs the strategies into the system. The ability of the system to solve problems improves with the addition of knowledge from different experts.

The ability to make the expertise of the reference staff available in a more consistent and timely fashion as well as free professionals from the burden of repetitive questions is the reason that expert systems have potential importance for reference services. There are a number of ways they can do this. One is by developing programs that can teach the reference process through machine simulation of the reference interview. Another is by providing basic assistance to users when personal help is not available. A third is by simulating the decision-making steps involved in negotiating a question and selecting appropriate source to answer it.

Many expert systems are now being applied in reference services. These include systems that recommend novels to readers, advise on dietary fiber, assist in choosing an index in biology or agriculture, translate natural language search inquiries into search statements that are acceptable for searching such things as the MEDLINE database, and instruct users in how to use Index Medicus for subject searches. In the 1980s interest in KBSs rapidly grew, and in 1988 these systems reached a development peak. Development and interest in these systems seems to have declined since then. Time, expertise, and resources to develop are required by a KBS. Then there is the uncertainty about the potential benefits as well. The advances in the technology of oonline public access catalogs and the World Wide Web might rekindle interest in knowledge based systems. There are now many Internet based expert systems that accommodate questions in plain English. The system offers answers to a query appear in about the same time it takes a search engine to do a search. It offers real-time reference assistance. Currently the performance of these systems is inconsistent but their refinement would allow expert systems to reach far beyond an individual library and benefit a far greater number of library customers.

The question of deprofessionalizing librarianship is going to come up as more expert systems are developed that can do more and more of the tasks that are now performed by reference staff. The desirability of removing reference staff from the intermediary role between user and information will be debated. The benefits of increasing access to reference expertise will also be debated. Determining just when such systems are appropriate in reference service will be something the reference staff will find necessary to determine. Sensitivity and objectivity as well as the needs of customers must be foremost in their minds while making these decisions.