Very large Bayesian multinets for text classification

  • Authors:
  • Mieczysław A. Kłopotek

  • Affiliations:
  • Institute of Computer Science, Polish Academy of Sciences, 01-237 Warsaw, ul. Ordona 21, Poland

  • Venue:
  • Future Generation Computer Systems
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper presents a newly developed algorithm learning very large tree-like Bayesian networks from data and exploits it to create a Bayesian multinet (BMN) classifier for natural language text documents.Results of empirical evaluation of this BMN classifier are presented. The study suggests that tree-like Bayesian networks are able to handle a classification task in 100000 variables with sufficient speed and accuracy.