Kalpani Anuradha Welivita

Ph.D. Student
School of Computer and Communication Sciences
École Polytechnique Fédérale de Lausanne (EPFL)

Postal Address
BC 144 (Bâtiment BC)
Station 14, CH-1015 Lausanne, Switzerland

Tel: +41 21 69 31203
Email: kalpani dot welivita at epfl dot ch

Research Interests

Dialogue Response Generation, Emotion Recognition, Natural Language Processing, Data Mining, Machine Learning, Human Computation, Human Computer Interaction

About My Research

My research lies mainly in the area of developing empathetic conversational agents. I analyze human-human conversations to discover strategies used by humans to empathize with others and utilize them in developing neuro-symbolic empathetic response generation models. My recent work also involves investigating psychotherapeutic discourse to embed empathetic chatbots with distress consolation and therapeutic capabilities.

Education

  • Ongoing: Ph.D. in Computer Science. School of Computer and Communication Sciences, École polytechnique fédérale de Lausanne (EPFL). Starting from September 2018.
  • B.S. in Engineering. Department of Computer Science and Engineering, University of Moratuwa, Sri Lanka. 2013 – 2017

Publications

  • Ekaterina Svikhnushina, Iuliana Voinea, Anuradha Welivita, and Pearl Pu. A Taxonomy of Empathetic Questions in Social Dialogs. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics, 2022. [access]
  • Anuradha Welivita and Pearl Pu. HEAL: A Knowledge Graph for Distress Management Conversations. In Proceedings of the 36th AAAI Conference on Artificial Intelligence, 2021. [introduction video] [full video] [poster]
  • Anuradha Welivita, Yubo Xie, and Pearl Pu. A Large-Scale Dataset for Empathetic Response Generation. In Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP), 2021. [access] [video]
  • Anuradha Welivita and Pearl Pu. A Taxonomy of Empathetic Response Intents in Human Social Conversations. In Proceedings of the 28th International Conference on Computational Linguistics (COLING), 2020. [access]