A paper about Few-Shot Emotion Recognition in Conversation accepted at EMNLP Conference

Affective computing 1200x630_1

Gaël Guibon is a post-doc researcher at Télécom Paris and SNCF. With Chloé ClavelMatthieu Labeau, professors at Télécom Paris, Hélène Flamein and Luce Lefeuvre, he wrote an article about Few-Shot Emotion Recognition in Conversation with Sequential Prototypical Networks that has been accepted at EMNLP, Conference on Empirical Methods in Natural Language Processing, Nov. 2021.

Gaël Guibon is a post-doctoral researcher in Natural Language Processing and Machine Learning at the Information Processing and Communications Laboratory (LTCI) of Télécom Paris and at the Innovation & Research Department from SNCF (French Railways). His research interests include emotion prediction and classification, emojis recommendation, machine learning from few examples (Few Shot Learning and Meta Learning), automatic French lexical evolution studies using machine learning and Old French dependency parsing and PoS tagging.

About the article

Several recent studies on dyadic human-human interactions have been done on conversations without specific business objectives. However, many companies might benefit from studies dedicated to more precise environments such as after sales services or customer satisfaction surveys. In this work, we place ourselves in the scope of a live chat customer service in which we want to detect emotions and their evolution in the conversation flow. This context leads to multiple challenges that range from exploiting restricted, small and mostly unlabeled datasets to finding and adapting methods for such context.We tackle these challenges by using Few-Shot Learning while making the hypothesis it can serve conversational emotion classification for different languages and sparse labels. We contribute by proposing a variation of Prototypical Networks for sequence labeling in conversation that we name ProtoSeq. We test this method on two datasets with different languages: daily conversations in English and customer service chat conversations in French. When applied to emotion classification in conversations, our method proved to be competitive even when compared to other ones.

Image source pch.vector/Freepik