Word Embedding Based Query Expansion.

Date of Submission

December 2018

Date of Award

Winter 12-12-2019

Institute Name (Publisher)

Indian Statistical Institute

Document Type

Master's Dissertation

Degree Name

Master of Technology

Subject Name

Computer Science

Department

Computer Vision and Pattern Recognition Unit (CVPR-Kolkata)

Supervisor

Mitra, Mandar (CVPR-Kolkata; ISI)

Abstract (Summary of the Work)

Continuous space word embeddings have received a great deal of attention in Information Retrieval for their ability to model term similarity and other relationships. We have studied the use of term relatedness in the context of query expansion for ad-hoc information retrieval. In our first approach we have proposed a time efficient retrieval algorithm for query expansion using pseudo-locally constrained word embeddings. In our second approach we have tried to present a learning approach that adaptively predicts the balance co-efficients between the original query model and the local and global expansion language models. In this approach we have also tried to predict the optimal number of expansion terms required for the local and global embedding based query expansion methods. In our third approach we have fused the results of query expansion based on local and global embeddings to have an improved performance over both the methods. In all the above approaches, we have performed our experiments on standard TREC ad-hoc data(Disk 4 and 5) with query sets TREC 6, 7, 8 and robust. Our first and third approaches have shown comparable performance with the state-of-the-art query expansion methods, based on word embeddings, but, our second approach has failed to perform in accordance to our hyposthesis.

Comments

ProQuest Collection ID: http://gateway.proquest.com/openurl?url_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:dissertation&res_dat=xri:pqm&rft_dat=xri:pqdiss:28843807

Control Number

ISI-DISS-2018-394

Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.

DOI

http://dspace.isical.ac.in:8080/jspui/handle/10263/6960

This document is currently not available here.

Share

COinS