Hybrid of SVM+PSO in CLIR Ranking using Word2Vector

Authors

  • Shweta pandey, Iti Mathur, Nisheeth Joshi

Abstract

Storing of information that is large in amount became possible after the realization of the importance of finding and archiving with the computer by the people.

In this paper we are using query in Hindi and using corpus dictionary the query is translated into English and related to the extracted keywords the relevant document or answer is classified using SVM and answer will be generated using PSO in English.

SVM  is used as a classifier which classifies the answers or documents .If condition doesn’t match the knowledge base gets updated and new updated answer will be generated .Finally using the PSO optimization the best answer is displayed after ranking. For feature extraction Word2Vector is applied. We have used PSO, SVM  and SVM-PSO  for the Implementation of CLIR and compare the results between them. The performance of the system thus improves 4 to 5 percent by SVM-PSO. FIRE 2011 ADHOC dataset is used for the performance evaluation of CLIR.

Keywords : CLIR, Ranking, PSO, SVM, Machine Learning

Published

2020-12-24

Issue

Section

Articles