Design for emotion detection of punjabi text using hybrid approach
Sheeba Grover, Amandeep Verma · 2016
In the era of autonomous technology, Emotion detection is a growing research area to develop an autonomous system for the detection of human emotions. There is the existence of various kinds of media on internet for articulation of human emotions like speech, images, face expressions, text etc. One of the ubiquitous sources of communication on internet is in the form of textual reviews, comments and other informational data. This textual data can be present in any of the language. The presence of data on internet is mainly in English language that lacks the other regional languages like Punjabi, Bengali, Telugu etc. In this paper, we are considering Punjabi language based textual data for emotion detection using hybrid concept of Keyword Based Approach and Machine Learning Approach. For keyword based approach, we have considered Rule Based Engine which detect whether the emotion is present in the input dataset or not. For machine learning approach, we considered Support Vector Machine (SVM) & Naive Bayes (NB) as classifier to detect the Ekman's six types of basic emotions (happy, fear, anger, sadness, disgust and surprise). For this experimentation, Punjabi textual dataset HC Corpora is used which consists of online Punjabi websites, Punjabi newspaper news, blogs etc. The proposed concept is implemented in python and dataset files are considered as the Unicode format.