Emotion Recognition in Conversation: Research Challenges, Negative Results, and Benchmarks
COM2 Level 4
Executive Classroom, COM2-04-02
Abstract:
Emotion is intrinsic to humans and they express emotion through language, be it verbal or physical. Thus, emotion emulation is a key part of human-like artificial intelligence (AI). Conversational emotion recognition (CER) is increasingly becoming popular as a new research frontier in natural language processing (NLP) due to its ability to mine opinions from the plethora of publicly available conversational data in platforms like Facebook, Youtube, Reddit, Twitter etc. Moreover, it has potential applications in health-care systems (as a tool for psychological analysis), human resources (evaluation of prospective employees) and much more. Besides, CER is also extremely important for generating empathetic dialogues which requires understanding the user's emotions. Catering to these needs calls for effective and scalable conversational emotion-recognition algorithms.
However, it is a strenuous problem to solve because of several research challenges. In this talk, I will discuss these challenges and as well as shed light on the recent advances in this research field.
Finally, I will also report a few negative results that we got while computationally addressing some of these challenges.
Biodata:
Soujanya Poria holds a Ph.D. degree in Computer Science from the University of Stirling, UK. He is a recipient of the prestigious early career research award called 'NTU Presidential Postdoctoral Fellowship' in 2018 which offers him a research grant worth US$150,000. Before taking up his presidential fellowship position at NTU, Soujanya was a scientist at the A*STAR and the Temasek Laboratory, NTU. He is also (co-)PI of multiple academic and industrial grants with the amount totaling to US$ 250,000. Soujanya has co-authored more than 60 papers, published in top-tier conferences and journals such as ACL, EMNLP, AAAI, NAACL, Neurocomputing, Computational Intelligence Magazine, etc.. He is also an adjunct faculty at Indraprastha Institute of Information Technology, Delhi, India and an adjunct scientist at A*STAR, Singapore. Soujanya served as a senior PC member at AAAI 2019, IJCAI 2019 and often serve as a PC member in reputed conferences such as ACL, EMNLP, IJCAI, NAACL. He was an area co-chair of the sentiment analysis track at NAACL 2019 and a publicity chair at *SEM 2019. Soujanya has given several invited talks at venues like CICLing 2018 which is a large international NLP conference. Soujanya has Google Scholar citations of more than 3900 and his h-index is 35. Three of Soujanya's papers are listed as Web of Science highly cited papers in the field of Computer Science. Recently, in an article published in the Journal of Information Sciences, Soujanya has been listed as one of the most prolific and impactful researchers from 2000 to 2016 in the field of sentiment analysis.