arXivDaily arXiv每日学术速递 周一至周五更新
看起来今天可以划水了
arXiv 2602.08709cs.CL
收藏

FactSim: 意见摘要中的事实核查

FactSim: Fact-Checking for Opinion Summarization

MercadoLibre São Paulo, Brazil(巴西圣保罗MercadoLibre);MercadoLibre Cordoba, Argentina(阿根廷 CordobaMercadoLibre)

Leandro Anghinoni, Jorge Sanchez

AI总结:

FactSim提出了一种全自动方法,用于评估意见摘要中生成内容的事实一致性,通过测量摘要声明与原始评论声明的相似性,提高事实核查的准确性。

摘要:

我们探索了在文本摘要任务中,特别是意见摘要领域,对生成式人工智能(GenAI)进行更全面和精确评估技术的必要性。传统方法通过自动指标比较生成摘要与意见文章(如产品评论)的摘要,但由于大型语言模型(LLM)带来的范式转变,这些方法存在局限性。本文通过提出一种全新的、全自动的方法来评估此类摘要的事实一致性来解决这些问题。该方法基于测量摘要中声明与原始评论中声明的相似性,测量生成摘要的覆盖范围和一致性。为此,我们依赖于从文本中提取事实评估的简单方法,然后进行比较和总结以形成合适的评分。我们证明了所提出的指标对相似声明赋予更高的分数,无论声明是否被否定、改写或扩展,并且与最先进的指标相比,该评分与人类判断有高度相关性。

英文摘要:

We explore the need for more comprehensive and precise evaluation techniques for generative artificial intelligence (GenAI) in text summarization tasks, specifically in the area of opinion summarization. Traditional methods, which leverage automated metrics to compare machine-generated summaries from a collection of opinion pieces, e.g. product reviews, have shown limitations due to the paradigm shift introduced by large language models (LLM). This paper addresses these shortcomings by proposing a novel, fully automated methodology for assessing the factual consistency of such summaries. The method is based on measuring the similarity between the claims in a given summary with those from the original reviews, measuring the coverage and consistency of the generated summary. To do so, we rely on a simple approach to extract factual assessment from texts that we then compare and summarize in a suitable score. We demonstrate that the proposed metric attributes higher scores to similar claims, regardless of whether the claim is negated, paraphrased, or expanded, and that the score has a high correlation to human judgment when compared to state-of-the-art metrics.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑