Document Type : Research Paper
1 Computer Applications, PSG College of Technology, Coimbatore, India
2 Applied Mathematics and Computational Sciences, PSG College of Technology, Coimbatore, India
The raw data obtained in the form of search results may be large for any particular problem, but is often a relatively small subset of the data that are relevant, and a search engine does not enable discovering the necessary subset of relevant text data in a large text collection. In this paper, a solution to a problem called conformity to truth, which studies how to find websites with the maximum amount of true facts, from a large amount of conflicting information on the user-defined topic, is proposed. Two algorithms called ParaSearch and FactFinder, which helps in identifying the best web links for searching general information and finding individual facts respectively are proposed. In ParaSearch, latent Dirichlet allocation (LDA) is used to identify the top 10 frequent terms using which we further construct a similarity matrix to identify the best web pages. In FactFinder, the usage of semantic processing is done to identify the best web pages, building upon the existing Page Rank Algorithm to further optimize the search results. The results prove that ParaSearch can identify web pages with the maximum number of facts conforming to the truth much better than popular search engines. The ambiguity of the individual facts is decreased to a great extent by using the FactFinder algorithm. Thus these algorithms will increase the accuracy of identifying possible web links for a given search word much better than most of the popular search engines.