Unstructured Text Enhanced Open-domain Dialogue System: A Systematic Survey

TL;DR

This survey comprehensively reviews unstructured text enhanced open-domain dialogue systems, analyzing their models, datasets, evaluation methods, and future trends to advance research in knowledge-integrated dialogue generation.

Contribution

It systematically categorizes UTEDS models, summarizes recent datasets and evaluation techniques, and discusses future directions, providing a structured overview of this emerging field.

Findings

Retrieval models include Fusion, Matching, and Ranking modules.

Generative models involve Dialogue and Knowledge Encoding, Knowledge Selection, Response Generation.

Current models show promising performance but face challenges in knowledge integration.

Abstract

Incorporating external knowledge into dialogue generation has been proven to benefit the performance of an open-domain Dialogue System (DS), such as generating informative or stylized responses, controlling conversation topics. In this article, we study the open-domain DS that uses unstructured text as external knowledge sources (\textbf{U}nstructured \textbf{T}ext \textbf{E}nhanced \textbf{D}ialogue \textbf{S}ystem, \textbf{UTEDS}). The existence of unstructured text entails distinctions between UTEDS and traditional data-driven DS and we aim to analyze these differences. We first give the definition of the UTEDS related concepts, then summarize the recently released datasets and models. We categorize UTEDS into Retrieval and Generative models and introduce them from the perspective of model components. The retrieval models consist of Fusion, Matching, and Ranking modules, while the generative models comprise Dialogue and Knowledge Encoding, Knowledge Selection, and Response Generation modules. We further summarize the evaluation methods utilized in UTEDS and analyze the current models' performance. At last, we discuss the future development trends of UTEDS, hoping to inspire new research in this field.