Emotion recognition from textual data -

Abstract

In today’s world and with the growth of internet community, textual data has proven to be the main tool of communication in human-machines or human-human interaction, where a large proportion of information is stored in textual form. This communication between humans and machines is constantly evolving towards the goal of making it as humane and real as possible. One way of humanizing such interaction is to provide a tool that can recognize the emotions present in the communication or the emotions of the involved users for the purpose of enriching the user experience. For example, by providing insights to the users for personal preferences and automated recommendations based on their emotional state. In this work, we introduce a method for accurate emotion recognition from text. We focus on extracting the six Ekman emotions present within the text. We propose a k-nearest neighbors (KNN) classifier based on syntactic and semantic analysis of the sentence. We start by generating an intermediate emotional data representation of the input sentence based on its structure. We then generalize this representation using WordNet and ConceptNet. Finally, a KNN classifier with a hand crafted similarity measurement equations is used to compare the input sentence with a set of reference emotion recognition rules extracted from the training set. We tested our classifier using different training sets. Our classifier outperformed the state-of-the-art machine learning and rule based classifiers and showed some encouraging results with an average F-Score of 76percent in textual emotion classification and with 85percent when excluding the emotion surprise from the dataset. The results also show the importance of syntactic and semantic analysis in emotion recognition from text.