Mitigating the Negative Impact of Over-association for Conversational Query Production

TL;DR

This work carefully analyzes the negative effects of the over-association phenomenon on pretrained Seq2seq query producers and proposes effective instance-level weighting strategies for training to mitigate these issues from multiple perspectives and shows that this model selects better concepts from dialogue histories and is 10 times more data efficient than the baseline.

Abstract

Conversational query generation aims at producing search queries from dialogue histories, which are then used to retrieve relevant knowledge from a search engine to help knowledge-based dialogue systems. Trained to maximize the likelihood of gold queries, previous models suffer from the data hunger issue, and they tend to both drop important concepts from dialogue histories and generate irrelevant concepts at inference time. We attribute these issues to the over-association phenomenon where a large number of gold queries are indirectly related to the dialogue topics, because annotators may unconsciously perform reasoning with their background knowledge when generating these gold queries. We carefully analyze the negative effects of this phenomenon on pretrained Seq2seq query producers and then propose effective instance-level weighting strategies for training to mitigate these issues from multiple perspectives. Experiments on two benchmarks, Wizard-of-Internet and DuSinc, show that our strategies effectively alleviate the negative effects and lead to significant performance gains (2%-5% across automatic metrics and human evaluation). Further analysis shows that our model selects better concepts from dialogue histories and is 10 times more data efficient than the baseline. The code is available at https://github.com/DeepLearnXMU/QG-OverAsso.

Figures (7)

• Figure 1: Two typical conversations for query production. Trained with over-association instances such as the upper example, the query producer fails to predict the target topic "angry birds" while generating irrelevant word "online".
• Figure 2: Analyses on the negative impacts of over-association.
• Figure 3: Predictive probability of QP(ALL) to gold queries in different subsets of WoI (left) and DuSinc (right) development set.
• Figure 4: Three methods to tackle side effects of over-association.
• Figure 5: Model performances on WoI test set with different number of instances for training.