Hi-index | 0.00 |
ANNOD is the name of a system developed at the National Library of Medicine (NLM), which implements a set of linguistic and empirical techniques that permit retrieval of natural language information in response to natural language queries. The system is based on Dr. Gerard Salton's SMART [1] document retrieval system and is presently implemented on a mini-computer as part of an Interactive TExt Management System, ITEMS.[2] Actual experience with retrieval of information from NLM's Hepatitis Knowledge Base (HKB), an encyclopedic hierarchical, full-text file, is presented. The techniques used in ANNOD include: automatic stemming of words, common word deletion, thesaurus expansion, a complex empirical matching (ranking) algorithm (similarity measure), and techniques expressly designed to permit rapid response in a mini-computer environment. Preliminary testing demonstrates high efficiency in identifying portions of a text which are relevant to users.