Contrastive Learning for Inference in Dialogue

Etsuko Ishii, Yan Xu, Bryan Wilie, Ziwei Ji, Holy Lovenia, Willy Chung, Pascale Fung

Submission Type: Regular Long Paper

Submission Track: Dialogue and Interactive Systems

Keywords: inference in dialogue, commonsense reasoning in dialogue, contrastive learning, semantic gap, dialogue comprehension, information gap, inductive reasoning

TL;DR: Assess the difficulty of inferences in dialogue based on their information gap and propose a contrastive learning approach to bridge the gap.

Abstract:

Inference, especially those derived from inductive processes, is a crucial component in our conversation to complement the information implicitly or explicitly conveyed by a speaker.
While recent large language models show remarkable advances in inference tasks, their performance in inductive reasoning, where not all information is present in the context, is far behind deductive reasoning.
In this paper, we analyze the behavior of the models based on the task difficulty defined by the semantic information gap -- which distinguishes inductive and deductive reasoning.
Our analysis reveals that the information gap between dialogue contexts and desired inferences renders the inductive inference process more challenging.
To mitigate this information gap, we investigate a contrastive learning approach by feeding negative samples.
Our experiments suggest negative samples help models understand what is wrong and improve their inference generations.

Submission Number: 659