Author(s)

A. Kasam & H.-C. Kwon

Abstract

Incorporating mechanism selection of a document, such as indexing and terms weighting techniques, alleviates the data retrieval complexity and plays a significant role in the performance of information retrieval and data mining policy. The goals of information retrieval (IR) present the user automatically and accurately with relevant data in answer to their queries in the form of documents. To achieve the goals, an intelligent approach is to use the vector space model, which models documents and queries as a number of indexed vectors in the terms space. Unfortunately during document searching, the performance is frequently interrupted by the appearance of numbers of uncertain fake documents (such as stop words or absent terms etc). In reality, the same problems may be solved in an intelligent way, so parallel research is needed. To meet the requirements of the performance complexities by taking account not only occurring terms but also absent terms in finding similarity of patterns among the document and the query vectors could be included. In this paper, a balanced term weighting scheme for a vector space model has been proposed as a novel method to improve the retrieval performance. Additionally, a dimension compression method has been included to minimize the computational cost. Under the proposed scheme, a certain amount of Meta data have been tested and the results strongly recommend to us that the proposed method is comparatively more effective than any other information retrieval (IR) system. It could also be applicable for academic and commercial purposes. Keywords: information retrieval (IR), terms-weighting, balance term weighting scheme (BTWS).

Keywords

information retrieval (IR), terms-weighting, balance term weighting scheme (BTWS).