Combined Approach for Page Ranking In Information Retrieval System Using Context and TF-IDF Weight

Abstract

Ranking in Information Retrieval System has been researched extensively in recent years. IR System is aimed at providing users the most relevant documents in minimum possible time. Therefore, providing fast and efficient result to the user is a major issue in determining the performance of the IR systems. Ranking of the pages is done after they have been indexed. Most of the existing architectures of IR system shows that they rely on keyword-based queries and the indexing is done based on the terms of the document and also consists an array of the posting lists, each posting list being associated with a term and containing the term along with the identifiers of the documents containing them. This paper proposes a ranking structure where ranking is done on the basis of a combination of the context of the document and on term basis. Context based indexing is considered in which all the available context along with the list of related terms of that context are stored. List of documents of particular contexts are stored in context repository. The indexing of the documents are done with respect to their context. To rank these documents a combination of context based weight (how much a document is relevant with a context) and TF-IDF weight (how much the user query is relevant to a document without considering context) are used. The ranking is done in decreasing order of their total weight.