Back to Search
Start Over
A contemporary combined approach for query expansion
- Source :
- Multimedia Tools and Applications. 81:35195-35221
- Publication Year :
- 2020
- Publisher :
- Springer Science and Business Media LLC, 2020.
-
Abstract
- The use of an automatic query expansion technique is to enhance the performance of the Information Retrieval System. Selecting the candidate terms for query expansion is an essential task to make query more precise to extract the most suitable documents. This paper provides a method to select the best terms for query enhancement. Firstly, the effect of abbreviation resolution, Lexical Variation, Synonyms, n-gram pseudo-relevance feedback, Co-occurrence method on baseline approaches of query expansion is analyzed.. In this work, we used the Okapi BM25 algorithm for ranking. We used Concept-based normalization to deal with concept terms. Here our results show the improvement in results than the baseline approach. A new combined technique that integrates lexical variation, synonyms, n-gram pseudo relevance feedback for query enhancement is proposed. For experimental purpose three English written datasets CACM, CISI, and TREC-3 is used. The obtained results show improvement in the performance of query expansion concerning mean average precision, F-measure, and precision-recall curve.
- Subjects :
- Normalization (statistics)
Information retrieval
Computer Networks and Communications
Computer science
InformationSystems_INFORMATIONSTORAGEANDRETRIEVAL
Relevance feedback
020207 software engineering
02 engineering and technology
Resolution (logic)
Combined approach
Ranking (information retrieval)
Query expansion
Okapi BM25
Ranking
Hardware and Architecture
0202 electrical engineering, electronic engineering, information engineering
Media Technology
Baseline (configuration management)
Software
Subjects
Details
- ISSN :
- 15737721 and 13807501
- Volume :
- 81
- Database :
- OpenAIRE
- Journal :
- Multimedia Tools and Applications
- Accession number :
- edsair.doi...........20a78319bf01cdb2888cdab5d47401cf