Implicit Emotion Detection from Text with Information Fusion

In this paper we have proposed an approach for emotion detection in implicittexts. We have introduced a combinational system based on three subsystems. Eachone analyzes input data from a different aspect and produces an emotion label asoutput. The first subsystem is a machine learning method. The second one is astatistical approach based on vector space model (VSM) and the last one is akeyword based subsystem with an information fusion component to aggregate thefinal output of main system. We analyzed the performance of our proposed systemon ISEAR dataset with seven emotions: anger, joy, sad, shame, fear, disgust andguilt. The results show that our combinational system outperforms each subsystemwith overall f-measure of 0.68 and at least up to 0.71 in terms of F1 in emotion levelexcept for anger. The overall performance of the main system is 9.13% better thanmachine learning module 16.6% better than VSM and 23% better than keyword-based.

Keywords:

Implicit Emotion Detection , Combinational System , Information Fusion , Machine Learning , Vector Space Model , Keyword Based