Detecting Response Generation Not Requiring Factual Judgment

TL;DR

This paper introduces a dataset and classification approach to identify dialogue responses that do not require factual correctness, enhancing dialogue systems' ability to balance attractiveness and factuality.

Contribution

It presents a new dataset (DDFC) and demonstrates effective classification models for detecting non-factual responses in dialogues.

Findings

Achieved up to 88% accuracy in classifying responses

Created a crowdsourced annotated dialogue dataset

Improved understanding of non-factual response detection

Abstract

With the remarkable development of large language models (LLMs), ensuring the factuality of output has become a challenge. However, having all the contents of the response with given knowledge or facts is not necessarily a good thing in dialogues. This study aimed to achieve both attractiveness and factuality in a dialogue response for which a task was set to predict sentences that do not require factual correctness judgment such as agreeing, or personal opinions/feelings. We created a dataset, dialogue dataset annotated with fact-check-needed label (DDFC), for this task via crowdsourcing, and classification tasks were performed on several models using this dataset. The model with the highest classification accuracy could yield about 88% accurate classification results.