Why are Sequence-to-Sequence Models So Dull? Understanding the Low-Diversity Problem of Chatbots
This addresses the issue of dull and repetitive chatbot responses, which is incremental as it builds on existing research by adding a new source and solutions.
The paper tackles the low-diversity problem in sequence-to-sequence models for chatbots, identifying model over-confidence as a key source and proposing directions like confidence penalties and label smoothing to address it.
Diversity is a long-studied topic in information retrieval that usually refers to the requirement that retrieved results should be non-repetitive and cover different aspects. In a conversational setting, an additional dimension of diversity matters: an engaging response generation system should be able to output responses that are diverse and interesting. Sequence-to-sequence (Seq2Seq) models have been shown to be very effective for response generation. However, dialogue responses generated by Seq2Seq models tend to have low diversity. In this paper, we review known sources and existing approaches to this low-diversity problem. We also identify a source of low diversity that has been little studied so far, namely model over-confidence. We sketch several directions for tackling model over-confidence and, hence, the low-diversity problem, including confidence penalties and label smoothing.