[论文解读] DiSCo: Making Absence Visible in Intelligent Summarization Interfaces

Eran Fainman, Hagit Ben Shoshan+2 more|arXiv (Cornell University)|Jan 12, 2026

Topic Modeling被引用 0

一句话总结

DiSCo 提出一个基于期望的框架，在 AI 生成的摘要中揭示缺失与偏离领域规范，相较于以存在性为主的摘要，能提升细节和决策支持。跨滑雪、海滩与市中心领域的用户研究显示感知有用性提升，尽管可读性略有下降。

ABSTRACT

Intelligent interfaces increasingly use large language models to summarize user-generated content, yet these summaries emphasize what is mentioned while overlooking what is missing. This presence bias can mislead users who rely on summaries to make decisions. We present Domain Informed Summarization through Contrast (DiSCo), an expectation-based computational approach that makes absences visible by comparing each entity's content with domain topical expectations captured in reference distributions of aspects typically discussed in comparable accommodations. This comparison identifies aspects that are either unusually emphasized or missing relative to domain norms and integrates them into the generated text. In a user study across three accommodation domains, namely ski, beach, and city center, DiSCo summaries were rated as more detailed and useful for decision making than baseline large language model summaries, although slightly harder to read. The findings show that modeling expectations reduces presence bias and improves both transparency and decision support in intelligent summarization interfaces.

研究动机与目标

• 识别智能界面中以存在为驱动的摘要的局限性，以及缺失信息如何具有诊断价值。
• Develop a domain-informed summarization framework (DiSCo) that surfaces deviations from domain topical expectations.
• Integrate expectation-based analysis with LLM-based generation to produce absence-aware summaries.
• Evaluate whether absence-aware summaries improve detail, relevance, and decision support for users.

提出的方法

• 从聚合的领域评审中构建领域话题期望。
• 利用基于LLM的推荐系统从评审中提取话题与情感信号。
• 通过 LvS 计算住宿话题偏差，识别过度呈现与缺失的主题。
• 将偏差信号融入结构化提示中，以用于基于LLM的摘要生成。
• 在受控研究中比较 DiSCo 摘要与基线仅存在性摘要。

实验结果

研究问题

• RQ1 缺失感知的摘要是否比仅存在的基线摘要在感知准确性和有用性方面有提升？
• RQ2 DiSCo 摘要是否提高细节与决策支持，并对可读性造成何种代价？
• RQ3 领域层面的期望如何影响用户对住宿评审的解读？

主要发现

• DiSCo 摘要在滑雪、海滩和市中心领域被评为更细致、对决策更有帮助。
• 与基线摘要相比，DiSCo 提高了感知的决策支持。
• 用户感觉 DiSCo 摘要的可读性略有下降，相较于基线。
• DiSCo 能有效揭示既异常频繁的主题，也有缺失但在领域中普遍存在的主题。
• 研究支持通过建模期望来降低存在偏见并提高智能摘要界面的透明度。