Abstract: 

With the enormous growth of social media, opinion and emotion mining has gained increasing importance to draw insights from social media communication for crucial decision making.  Emojis have become ubiquitous, and people are doing extensive usage of emojis over social media to convey feelings and emotions. 'Face with Tears of Joy' emoji 😂 was declared Oxford word of the year in 2015, so extensive emojis presence eases the challenge to detect emotions in code mixed textual contents. We aim to analyze whether emojis can predict user emotions in the Pakistani social media context as most of the social media content of the public here is in code mixed English-Roman Urdu textual form. We are building an annotated corpus from code-mixed content of public WhatsApp groups in Pakistan for emotional analysis. A review of the latest emoji-emotion association approaches has been done and we are aiming to utilize Machine Learning classifiers (SVM, MNB) and a deep learning technique to build an emotion detection system for analyzing how well the emojis can forecast emotions in text and whether emojis reinforce textual emotion or serves as a replacement of emotion words for Pakistani users. We aim to take advantage of this system for predicting potential violent behavior in public social media groups.

Committee: 

• Dr. Agha Ali Raza (Advisor)
• Dr. Maryam Mustafa (Evaluator)

Meeting Link: https://zoom.us/j/97708644229?pwd=Z3JlaWU4ditBdG1MS05RTXo3SmhuUT09

Meeting ID: 977 0864 4229
Passcode: 45JjXN