Theoretical foundation of relevance frequency for text categorization

Amir Ahmad, Mainuddin, Santosh Kumar Ray

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

In text mining finding the significance of each term is a text is an important problem. In a two-class setting, relevance frequency is used to compute the discriminating power of a term for text categorization. Relevance frequency has shown better performance than other term weighting methods on number of datasets. However, relevance frequency is based on intuitive considerations. In this paper, we develop a probabilistic model to explain relevance frequency. The model provides a theoretical foundation for relevance frequency.

Original languageEnglish
Title of host publication2017 International Conference on Intelligent Computing, Instrumentation and Control Technologies, ICICICT 2017
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages122-125
Number of pages4
ISBN (Electronic)9781509061068
DOIs
Publication statusPublished - Apr 19 2018
Event2017 International Conference on Intelligent Computing, Instrumentation and Control Technologies, ICICICT 2017 - Kannur, India
Duration: Jul 6 2017Jul 7 2017

Publication series

Name2017 International Conference on Intelligent Computing, Instrumentation and Control Technologies, ICICICT 2017
Volume2018-January

Conference

Conference2017 International Conference on Intelligent Computing, Instrumentation and Control Technologies, ICICICT 2017
Country/TerritoryIndia
CityKannur
Period7/6/177/7/17

Keywords

  • Text categorization
  • discriminating power
  • inverse document frequency
  • relevance frequency

ASJC Scopus subject areas

  • Artificial Intelligence
  • Computer Networks and Communications
  • Hardware and Architecture
  • Control and Optimization
  • Instrumentation

Fingerprint

Dive into the research topics of 'Theoretical foundation of relevance frequency for text categorization'. Together they form a unique fingerprint.

Cite this