As far as Internet grows, the problem of pertinent search raises over and over again: as new indexing/search technique appears on the market and, seemingly, covers all the needs, several years of Internet growth raise the problem again. Currently, there is a great deal of interest in solving the problem of term ambiguity (How many meanings do such words as 'line' and 'square' have? At least, 36 and 24, respectively). A wide variety of approaches, such as using thesauri for modifying documents and queries with semantically similar terms and automatic term disambiguation into term senses have been investigated since early 1980's. There are not too many successful solutions, however.
The main reasons why existing search systems often produce irrelevant results are:
- Existing search engines rarely use thesauri and knowledge about the lexical structure of both document and user queries
- Ambiguity problem, caused by polysemy of the natural language
After years of experiments Novosoft Knowledge Management team came to conclusion that successful solution of the ambiguity problem is not a fully automated query-expanding technique, but a semi-automated system involving the user in the process of query construction. The way we follow now is the development of the search system that will interact with the user and make proper query-expansion/enforcement, based on lexical structures found in both user query and lexical databases.
Project goal is to develop a search-engine based on newest linguistic and ontological approaches and our proprietory technologies.
- Creation of user-friendly meta-search systems
- Relevant search in large text corpora