Question Answering (QA) System

  • Recently, I am working on learning sentence representation for natural language processing (NLP) tasks, including QA—published in ACL 2020.
  • We present a graph neural network-based model that can detect supporting sentence for machine-reading question answering—published in LREC 2020.
  • We propose a hierarchical model for understanding lengthy text for QA. In addition, we develop a latent clustering method that analyses and uses topic information in the target dataset as additional information—published in NAALC 2018, CIKM 2019.

NLP & Multimodal Information

  • We present a novel attention mechanism over multimodal information. It iteratively hops over each modality to aggregate salient information for speech emotion recognition—published in ICASSP 2019, ICASSP 2020, and Interspeech 2020.
  • We show that usage of multimodal information using deep neural network-based model significantly improve the performance for speech emotion recognition task—published in IEEE SLT 2018.

NLP for Social Good

  • We are trying to develop an algorithm that can be proactively used to prevent text-related problems in our society. Our first effort begin with developing a model that can detect the abusive language on Twitter—published in EMNLP workshop 2018. We further present research that can detect misleading news headlines—published in AAAI 2019, Disinformation, Misinformation, and Fake News in Social Media-Emerging Research Challenges and Opportunities, Springer 2020, and IEEE Access 2021.