TEXT BASED EMOTION DETECTION

Interaction of human with computer is very interesting and most famous area of research these days because the word is getting modern and digitize. This needs the digital systems to imitate the human behaviour correctly. Emotion is a particular part of human behaviour which plays an important role while interacting with computer, the computer interfaces need to detect the emotion of the users in order to build a truly intelligent behaviour. Every day, massive amount of textual data is gathered into internet such as blogs, social media etc. With the rapid growth of web application, most of documents are available on web in the form of text. So, detecting affects from text is a vital issue. Hence, attitude detection from text is important in many areas such as decision making, human computer interaction etc. Work done in this field is very less as compare to other fields. Therefore, it broadens our scope in the field of attitude detection. In this paper, we propose a hybrid model that incorporates natural language processing technique, including keyword-based and machine learning-based emotion classification from textual data at sentence level. Using proposed algorithm, one can calculate the affect vector of sentence by affect vector of word. Then based on affect vector categorize the sentence into appropriate affect class.