Sentiment analysis of Thai children stories

395

Views

0

Downloads

Pasupa, Kitsuchart, Netisopakul, Ponrudee and Lertsuksakda, Rathawut (2016) Sentiment analysis of Thai children stories Artificial Life and Robotics, 21 (3)., 357-364.

Abstract

Sentiment Text Tagging System (STTS) with Thai sentiment resource has been developed and used to tag emotions directly to words and sentences in Thai children stories. The Thai sentiment resource, developed from SenticNet2 resource, groups emotions into four independent but concomitant dimensions: pleasantness, attention, sensitivity and aptitude. The measure of each dimension is called a sentic value of that dimension. Thai sentiment resource stores each word’s sentic value and polarity value, a value calculated from the sentic value, in the form of floating point number. The resource was constructed from bi-directional translation of 14,244 English terms in SenticNet2 into 16,584 Thai terms. The main purpose of this study was to implement a sentiment analysis of Thai children stories system with support vector machine using a set of proposed discriminating features for classifying emotions. It was found that the system can achieve 75.67 % of accuracy.

Item Type:

Article

Identification Number (DOI):

Deposited by:

ระบบ อัตโนมัติ

Date Deposited:

2021-09-06 03:38:08

Last Modified:

2021-10-05 12:42:15

Impact and Interest:

Statistics