The Automation of Controlled Vocabulary Subject Indexing of Medical Journal Articles Some General Considerations concerning Indexing for Bibliographic Databases Information Retrieval Research and Subject Indexing the Subject Indexing Process: a Description of Key Elements and Procedures for Medline Indexing

The purpose of this article is to investigate the possibility of the automation of sophisticated subject indexing of medical journal articles. Approaches to subject descriptor assignment in information retrieval research are usually either based upon the manual descriptors included in the database or the attempted generation of alternative search parameters, using statistical, probabilistic or natural language methods or concept networks. A description of the principles of the Medline indexing system is presented, followed by a summary of the outcome of a pilot project, based upon the Amed database. The results suggest that a more extended study, based upon Medline, should encompass various components: 1. Extraction of "concept strings" from titles and abstracts of records, based upon a detailed analysis of linguistic features characteristic of the domain. 2. Mapping rules to associate the concept strings with entries in the Unified Medical Language System (UMLS, a consolidation of various medical vocabularies produced by the National Library of Medicine). 3. Coordination of the descriptors, utilising features of the Medline indexing system and a feedback mechanism relating to the original input. The emphasis should be on system manipulation of data, based upon input, available resources and specifically designed rules, avoiding any implication of system "understanding". Introduction The inclusion of subject indexing terms in the records of bibliographic databases is a common practice. While the sophistication of indexing systems may vary considerably, the objective of this activity is to facilitate effective and comprehensive information retrieval from the databases. The purpose of this article is to consider various aspects relevant to the automation of sophisticated subject indexing. While the utility of such subject indexing may be contentious, the assumption underlying the article is that it does have a useful, even essential, function. A brief overview of some general considerations regarding databases and indexing is presented. Reference is made especially to Medline, one of the best known databases, produced by the National Library of Medicine (NLM), as an exemplar of a database utilising a sophisticated subject indexing system. Next, some of the main lines of information retrieval research are briefly considered with respect to how this research has related to subject indexing. This is followed by a description of the manual subject indexing process, conducted by trained indexers, for the Medline database. The results of a pilot project investigating some aspects of the automation of the indexing process are then presented. These suggest that the possibility of …