A Free-text Database Retrieval System Incorporating The Semantic Network Wordnet

Searches of free text databases depend on the keywords used in the request for information. If keywords are extracted from the user's query and no use is made of synonyms, relevant data in the database may be overlooked. The same can be the case if class-subclass, and part-whole information is ignored. The semantic network WordNet that has been developed a Princeton University encodes a very rich store of semantic information concerning relationships between words that are used constantly in any language understanding. These include subordinate-superordinate and part-whole information, as well as refinements to them, restrictions of relationships to appropriate contexts and mappings between words of different categories (nouns, adjectives, verbs). It is envisioned that uses of the relationships encoded in WordNet can allow the computer to make deductions allowing it to determine that many texts are relevant to the needs that a user expressed in quite different terms. Netrologic proposes to develop a system to categorize articles based on concepts encoded in WordNet, and to create a free text database interface to search documents categorized in this way. ANTICIPATED BENEFITS: The efficient determination of all documents in a database relevant to a user's needs is becoming increasingly important, as larger and larger free text databases are coming in line. A system that contains an embedded semantic network can be an invaluable aid to everyone in both government and the general public who will be making extensive uses of large databases.