ABSTRACT
Information retrieval can be regarded as a process through which a gap in an individual’s cognitive map is filled by information, or knowledge, retrieved from an external source of information, or knowledge. Viewed in this way, it is clear that artificial intelligence (AI) approaches could be used to fill this gap in the individual’s cognitive map, and various approaches along these lines have been attempted. During the 1980s, the expert-system paradigm dominated AI research, particularly in medicine, but in many other fields as well. Despite this research interest, the main information retrieval systems used operationally to retrieve information for biomedical researchers have not incorporated AI approaches, nor is ranked document retrieval used by these systems, for example, PubMed. In the past few years, there has been much interest in Semantic Web technologies in the biomedical community, and in 2009, the National Institutes of Health (NIH) funded two companion stimulus grant projects for semantic repositories and search engines: eagle-i and VIVO [eagle-i 2012; VIVO 2012]. One research area that emerged from expert-system research was uncertainty and AI [Association 2012]. Expert systems are generally written using a rule-based formalism that captures the reasoning process of human experts in a given narrow domain. Originally, such rules were written either deterministically or with ad hoc confidence factors. The field of uncertainty and AI emerged as researchers became interested in providing better formalisms for representing uncertainty. More will be said about this later in the chapter. One formalism that emerged was the Bayesian belief network.
