Detecting the links between utterances in character based multi-party conversation by machine learning with meta-information

  • Authors:

  • Junpei Nakamura;Yasuhiro Tajima;Yoshiyuki Kotani

  • Affiliations:

  • Department of Electronic and Information Engineering, Tokyo University of Agriculture and Technology, Koganei, Tokyo, Japan;Department of Electronic and Information Engineering, Tokyo University of Agriculture and Technology, Koganei, Tokyo, Japan;Department of Electronic and Information Engineering, Tokyo University of Agriculture and Technology, Koganei, Tokyo, Japan

  • Venue:
  • CEA'07 Proceedings of the 2007 annual Conference on International Conference on Computer Engineering and Applications
  • Year:
  • 2007

Quantified Score

Hi-index 0.00

Visualization

Abstract

It is important to find out interactive links between pairs of utterances in multi-party conversation like an online chat. Though the usage of linguistic information is necessary to do this, we showed the better performance to this criterion by using physical meta-information that consists of the number of conversation members, the distance between utterances, and the frequency of individual utterance. The result of the examination of Support Vector Machine (SVM) learning showed the accuracy is 81.3%, the precision is 74.3% and the recall is 77.9% for link between same perosn's utterances, and the accuracy is 80.3%, the precision is 71.1% and the recall is 66.8% for link between others' utterances. The result of the examination without meta-information showed the accuracy is 63.9%, the precision is 50.9%, the recall is 53.3% for same person's utterances, and the accuracy is 79.5%, the precision is 74.7% and the recall is 57.0% for others'. These results showed we could find new links by using meta-information.