Maximum Entropy Principle based Document Ranking with Term Selection Analysis for Cross-Lingual Information Retrieval

Retrieval of proper information from the collection of available data on Internet in a cross-lingual platform is a tedious task. For the retrieval of information, the user specifies the required data in the way of query. In some cases, the query could not clearly define the required information specifically owing to the ambiguities and is solved by the use of Google translator. This paper presents a new Tamil-English cross lingual information retrieval (CLIR) where the queries in Tamil language are used to retrieve related documents from English language. The proposed model uses Maximum Entropy Principle based Document Ranking with Term Selection Analysis (MEPDR-TSA) for CLIR. The presented MEPDR model distributes the term frequency weights between the documents subjected to a provided query. In TSA, the term selection value is used to select the retrieved documents from each query. In addition, Google translator is used to translating the query into English and then again translate the retrieved English documents into Tamil language. The performance of the MEPDR-TSA model has been tested against several Tamil language based queries and the results are investigated interms of precision, recall, and F-score.