An integration of triple parser with predicate based approach for search engine development

Abstract

The rapid development of modern technology has resulted in large amount of electronically available information in articles and patents. The search engines are programs that searches the documents in a database correspond to queries specified by the user. Searching by using keywords for millions of documents will not be precise and thus retrieve incomplete information. This makes it difficult for the user to detect relevant documents manually from large document collection. The keyword based approach uses the vector space model for representing both query and documents as vectors. Terms with partial matching are not associated in this method. Predicate based approach is the method to provide the user with a more precise and complete information with a single query using predicates for both query and document representation. Predicates are also called triples that contain complex data structures and are more structured information. Predicate based search engine is done by combining several methods such as predicate based vector space model, query-document similarity function with adjusted TF-IDF and boost function. The goal is to find feasible methods to develop the predicate parser to generate a well structured text. The boost function is integrated with similarity measure to generate a rich and sophisticated information search. The proper order of relevant documents will considerably improve the performance.