PMIScore: An Unsupervised Approach to Quantify Dialogue Engagement

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Guo, Yongkang, Huang, Zhihuan, Kong, Yuqing
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915862481469440
author Guo, Yongkang
Huang, Zhihuan
Kong, Yuqing
author_facet Guo, Yongkang
Huang, Zhihuan
Kong, Yuqing
contents High dialogue engagement is a crucial indicator of an effective conversation. A reliable measure of engagement could help benchmark large language models, enhance the effectiveness of human-computer interactions, or improve personal communication skills. However, quantifying engagement is challenging, since it is subjective and lacks a "gold standard". This paper proposes PMIScore, an efficient unsupervised approach to quantify dialogue engagement. It uses pointwise mutual information (PMI), which is the probability of generating a response conditioning on the conversation history. Thus, PMIScore offers a clear interpretation of engagement. As directly computing PMI is intractable due to the complexity of dialogues, PMIScore learned it through a dual form of divergence. The algorithm includes generating positive and negative dialogue pairs, extracting embeddings by large language models (LLMs), and training a small neural network using a mutual information loss function. We validated PMIScore on both synthetic and real-world datasets. Our results demonstrate the effectiveness of PMIScore in PMI estimation and the reasonableness of the PMI metric itself.
format Preprint
id arxiv_https___arxiv_org_abs_2603_13796
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle PMIScore: An Unsupervised Approach to Quantify Dialogue Engagement
Guo, Yongkang
Huang, Zhihuan
Kong, Yuqing
Computation and Language
High dialogue engagement is a crucial indicator of an effective conversation. A reliable measure of engagement could help benchmark large language models, enhance the effectiveness of human-computer interactions, or improve personal communication skills. However, quantifying engagement is challenging, since it is subjective and lacks a "gold standard". This paper proposes PMIScore, an efficient unsupervised approach to quantify dialogue engagement. It uses pointwise mutual information (PMI), which is the probability of generating a response conditioning on the conversation history. Thus, PMIScore offers a clear interpretation of engagement. As directly computing PMI is intractable due to the complexity of dialogues, PMIScore learned it through a dual form of divergence. The algorithm includes generating positive and negative dialogue pairs, extracting embeddings by large language models (LLMs), and training a small neural network using a mutual information loss function. We validated PMIScore on both synthetic and real-world datasets. Our results demonstrate the effectiveness of PMIScore in PMI estimation and the reasonableness of the PMI metric itself.
title PMIScore: An Unsupervised Approach to Quantify Dialogue Engagement
topic Computation and Language
url https://arxiv.org/abs/2603.13796