|Goals:||Create a taxonomy of English interjections with exhaustive lists of examples that could serve for enhancing the naturalness of chatbot’s responses.|
|Assistant:||Ekaterina Svikhnushina (ekaterina DOT svikhnushina AT epfl DOT ch)|
|Keywords:||Interjections; discourse analysis; taxonomy; natural language processing|
Interjections are words and expressions that people use to communicate sudden reactions, feelings, and emotions. Most of the time we use interjections unconsciously and perceive them as an integral part of a human conversation. As naturalness is an important aspect for open-domain conversational agents, active research is being conducted on the role of interjections for chatbots. Earlier studies demonstrated that chatbots are evaluated as more natural and engaging when they use interjections in their responses (Marge, 2010; Cohn, 2019). However, previous works operated with limited lists of interjections and used custom heuristic rules and classifications to introduce them into agent’s utterances. The aim of this project is to create a robust taxonomy of interjections enriched with numerous examples so that it could be used reliably to enhance naturalness of chatbot’s responses.
The student is expected to:
|Related Skills:||Strong analytical skills; background and/or interest in discourse analysis; familiarity with natural language processing|
|Suitable for:||Master student. Interested student should contact Ekaterina Svikhnushina (ekaterina DOT svikhnushina AT epfl DOT ch) and Pearl Pu (pearl DOT pu AT epfl DOT ch) along with a copy of your CV.|