Knowledge-based (expert) systems

In Uncategorized

General Definition

 

Knowledge-based systems (also called knowledge-based expert systems) aim to simulate human expertise in the design of non-human systems. The design of these kinds of systems is generally considered as a subset of artificial intelligence, a discipline that attempts to “emulate the problem process of the expert(s) whose knowledge was used in the development of the system.” (Williams and Sochat, para. 4) Knowledge-based systems have applications in a variety of disciplines, including engineering, strategic planning, medicine, navigation, and even the game of chess. (Richardson)

 

Knowledge-based systems in Reference

 

Within the library environment, knowledge-based systems have several applications in the area of cataloging, collections development, and indexing. Significant attention has also been given to knowledge-based systems in the reference environment, where such a system attempts to draw upon and simulate the expertise of a reference librarian.

 

Components of Knowledge-based Systems

 

Knowledge-based systems can vary to some degree in their components, but the essential features are considered to be a knowledge base, an inference engine, and a user interface. Other features of knowledge-based systems may include an explanation capability, an interface with other systems, and the ability to update its own knowledge base.

 

In the reference environment, the knowledge base is considered to be the reference sources, strategies for when to use certain types of sources, and a “deeper knowledge of interviewing users and the structure of knowledge within a particular domain.” (Williams & Sochats, para. 28). Knowledge can be represented in the form of rules, frames, examples, groups, probabilities and logic. (Richardson, 185-187)

 

An inference engine built in to the system attempts to provide a link between the information that a user has entered and the appropriate match within the knowledge base. The inference engine may employ different methods to arrive at a conclusion or goal. Backward chaining begins with the goal, and works backward from this goal to match the information provided in the premise. Forward chaining, considered more “data driven” (Richardson, 188), starts from the conditions presented, then attempts to match these conditions with knowledge in the base. If a match is found, the goal is achieved. Chowdhury notes that backward chaining may be a more useful method when a problem has numerous conditions, but few conclusions, while forward chaining may be useful when there are few conditions but many conclusions. (Chowdhury, 324) Knowledge-based systems may also use inductive reasoning, drawing upon examples to reach a conclusion, or inheritance, in which a “child” inherits certain characteristics of a “parent.” (Richardson, 188-189)

 

The user interface is a key element of the knowledge-based system, and a successful system will be able to communicate with human users in a natural way. Natural language processing therefore is an important part of the system, as the needs of the user must be translated from natural language into a form that a computer can understand, and in providing a recommendation for a course of action, the computer must communicate back to the user in a way that the user can understand.

 

Knowledge-based systems have the potential to allow a greater efficiency in the library environment, as they may provide assistance to users when no human "expert" is available. Within the reference environment, the development of knowledge-based systems will benefit from continued research in to the nature of the reference transaction. Such research will inform decisions about what sorts of knowledge will be represented in the knowledge-base, how this knowledge will be represented and communicated, and how different users may utilize these systems.

 

Sources

 

Chowdhury, G. G. (1999). Introduction to Modern Information Retrieval. London: Library Association Publishing.

 

Feathers, J., & Sturges, P. (Eds.). (2003). International Encyclopedia of Information and Library Science (2nd Edition). London and New York: Routledge.

 

Richardson, J. V., Jr. (1995). Knowledge-based Systems for General Reference Work: Applications, Problems, and Progress. San Diego: Academic Press.

 

Sochats, K., & WIlliams, J.G. (1996). Application of Expert Agents/Assistants in Library and Information Systems. DESIDOC Bulletin of Information Technology, 16(4), 19-32. Retrieved November 21, 2006 from http://ltl13.exp.sis.pitt.edu/Website/Webresume/ExpertAgentsPaper/Expert.htm

 

Elizabeth H. Bornheimer