Using Multi-Encoder Fusion Strategies to Improve Personalized Response Selection

TL;DR

This paper introduces fusion strategies that integrate persona, emotion, and entailment information to enhance personalized response selection, significantly improving accuracy and achieving state-of-the-art results on the Persona-Chat dataset.

Contribution

It proposes novel fusion strategies and a concept-flow encoding for BERT-based models to better capture interaction between persona, emotion, and context in response selection.

Findings

Incorporating emotion and entailment improves response selection accuracy.

Fusion strategies combined with concept-flow encoding outperform previous methods.

Achieved over 2.3% higher hits@1 on Persona-Chat dataset.

Abstract

Personalized response selection systems are generally grounded on persona. However, there exists a co-relation between persona and empathy, which is not explored well in these systems. Also, faithfulness to the conversation context plunges when a contradictory or an off-topic response is selected. This paper attempts to address these issues by proposing a suite of fusion strategies that capture the interaction between persona, emotion, and entailment information of the utterances. Ablation studies on the Persona-Chat dataset show that incorporating emotion and entailment improves the accuracy of response selection. We combine our fusion strategies and concept-flow encoding to train a BERT-based model which outperforms the previous methods by margins larger than 2.3 % on original personas and 1.9 % on revised personas in terms of hits@1 (top-1 accuracy), achieving a new state-of-the-art performance on the Persona-Chat dataset.