Text Classification based Behavioural Analysis of WhatsApp Chats

WhatsApp is used by millions of users to express emotions and share feelings. The model is presented in this paper aims to perform sentimental and emotional analysis using textual messages and emojis used in WhatsApp chats. Code switching, which is quite prevalent over online conversations, is handled by the model by unifying and converting all the texts to a standard form. For each subject, multiple chats are taken; translated and using a neural network, each sentence and emoji is scored in a dimensional form. The composition of the emotions expressed by the subject (out of Happy, Sad, Bored, Fear, Anger and Excitement) are defined. The scores are added up for each subject. Throughout the analysis, the behavioral traits are extracted. It is determined that, if the subject likes to use emojis and if they use it as a replacement for words or as an add-on to express their emotions better. It is also observed that if the subject behaves differently on text according to the person in front of them with regard to these emotions and finally, if the subject is an introvert or extrovert.