MECHANIZED INDEXING STUDIES OF MSD TOXICITY. PART I AND PART II,
Abstract
Mechanized indexing possibilities are being studied by searching for computer rules which will duplicate indexing done by subject specialists for a pharmaceutical retrieval system, Toxicity (including side effects) is the first term studies. A random hundred documents, containing thirteen toxicity papers, is the first sample. Computer-generated word frequency lists for each paper aid the work. The much discussed keyword approach is tried first. A thesaurus group of likely toxicity keywords is derived from the retrieval system's indexing guide, a medical dictionary, and the papers in the sample. Frequencies and document places (occurrence in titles, summaries, headings, etc..) of keywords, and differing weights for keywords, are used to try to reduce keyword overassigning of toxicity. Frequencies do not appear to help. Weighting helps some. Occurrence in summaries helps more. No keywords occur in titles. The Appendices (bound separately) include a bibliography of all sample papers containing keywords. Each reference is accompanied by a brief summary of the paper, and brief summaries of the passages containing keywords. Another Appendix gives for each paper its word length and the identity, document places, and frequences of the keywords it contains. Another Appendix contains the full texts or excerpts of the five toxicity papers which appear to contain no keywords. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1964
- Accession Number
- AD0436523
Entities
People
- John O'conner
Organizations
- VINITI