Question Classification Based on Improved TFIDF Algorithm

DOI

10.2991/caai-17.2017.80How to use a DOI?

Keywords

TFIDF; mutual information; information entropy; space vector model; questioning classification

Abstract

The feature weight calculation is an important part of question classification, feature weights directly affect the accuracy of the classification results, the traditional TFIDF algorithm is widely applied in the field, but it ignores the relationship between features and classes. Therefore, this paper introduces information entropy and mutual information to improve the traditional TFIDF, I convert the computed weight results into a space vector model format, and use the support vector machine (SVM) for question classification. Finally, I use TFIDF, TFIDF-MI and TFIDF-MI-E these three methods to calculate the feature weight, and compare the results of the classification experiments. The experimental results show that the accuracy, recall and F value of TFIDF-MI-E algorithm are higher than those of TFIDF and TFIDF-MI two algorithm.

Copyright

© 2017, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Jing Gao
AU  - Cui-xiao Zhang
AU  - Zhiqiang Wang
AU  - Guangzhen Zhao
AU  - Xuan Li
PY  - 2017/06
DA  - 2017/06
TI  - Question Classification Based on Improved TFIDF Algorithm
BT  - Proceedings of the 2017 2nd International Conference on Control, Automation and Artificial Intelligence (CAAI 2017)
PB  - Atlantis Press
SP  - 354
EP  - 357
SN  - 1951-6851
UR  - https://doi.org/10.2991/caai-17.2017.80
DO  - 10.2991/caai-17.2017.80
ID  - Gao2017/06
ER  -