MedIndEx System: Research on Interactive Knowledge-Based Indexing and Knowledge Management

摘要

The MedIndEx (Medical Indexing Expert) System is an interactive, knowledge-based prototype for computer-assisted indexing of the biomedical literature being developed at the Lister Hill National Center for Biomedical Communications (LHNCBC), a research division of the National Library of Medicine (NLM). Principal areas of research are knowledge representation and indexing for information retrieval. The system uses an experimental frame language featuring inheritance and procedural attachments, and is designed to interact with trained MEDLINE indexers by prompting them to enter MeSH (Medical Subject Headings) terms as slot fillers in completing document-specific indexing frames derived from knowledge base frames corresponding to MeSH concepts. The system provides many types of interactive assistance, including prescriptive and suggestive indexing aids. Based on information in these indexing frames, the system generates conventional indexing terms, consisting of MeSH descriptors with appended topical qualifiers. The report updates previous NTIS reports PB87-175790/AS and PB88-243902/AS. Since the latter, a knowledge base manager has been added to the system. This is a facility for assisting knowledge engineers in developing MedKB, the system's knowledge base, ensuring both consistency and proper syntax. Details of this and other recent developments are in the first part of the report, Introduction and Summary of Developments. Remaining three parts are MedIndEx System Design Document; MedIndEx User Manual: Indexing; and References.