Grounding in social media: An approach to building a chit-chat dialogue model

TL;DR

This paper proposes a social media grounded approach to enhance open-domain dialogue systems by mimicking human responses through casual Reddit interactions, improving conversational richness and relevance.

Contribution

It introduces a joint retriever-generator model that leverages Reddit comments to provide external context, broadening beyond traditional structured knowledge sources.

Findings

Improved response relevance in automatic evaluations.

Enhanced human-like conversational quality.

Effective use of social media data for dialogue modeling.

Abstract

Building open-domain dialogue systems capable of rich human-like conversational ability is one of the fundamental challenges in language generation. However, even with recent advancements in the field, existing open-domain generative models fail to capture and utilize external knowledge, leading to repetitive or generic responses to unseen utterances. Current work on knowledge-grounded dialogue generation primarily focuses on persona incorporation or searching a fact-based structured knowledge source such as Wikipedia. Our method takes a broader and simpler approach, which aims to improve the raw conversation ability of the system by mimicking the human response behavior through casual interactions found on social media. Utilizing a joint retriever-generator setup, the model queries a large set of filtered comment data from Reddit to act as additional context for the seq2seq generator. Automatic and human evaluations on open-domain dialogue datasets demonstrate the effectiveness of our approach.