Emotion Detection in Unfix-length-Context Conversation
This addresses emotion recognition in multi-turn dialogues, which is an incremental improvement for conversational AI applications.
The paper tackles emotion detection in conversations by using variable-length context windows for different utterances, achieving performance improvements over strong baselines on three public datasets.
We leverage different context windows when predicting the emotion of different utterances. New modules are included to realize variable-length context: 1) two speaker-aware units, which explicitly model inner- and inter-speaker dependencies to form distilled conversational context, and 2) a top-k normalization layer, which determines the most proper context windows from the conversational context to predict emotion. Experiments and ablation studies show that our approach outperforms several strong baselines on three public datasets.