GlobeSumm: A Challenging Benchmark Towards Unifying Multi-lingual, Cross-lingual and Multi-document News Summarization

TL;DR

The GLOBESUMM dataset is meticulously constructed by first collecting a wealth of multilingual news reports and restructuring them into event-centric format and the method of protocol-guided prompting for high-quality and cost-effective reference annotation is introduced.

Abstract

News summarization in today's global scene can be daunting with its flood of multilingual content and varied viewpoints from different sources. However, current studies often neglect such real-world scenarios as they tend to focus solely on either single-language or single-document tasks. To bridge this gap, we aim to unify Multi-lingual, Cross-lingual and Multi-document Summarization into a novel task, i.e., MCMS, which encapsulates the real-world requirements all-in-one. Nevertheless, the lack of a benchmark inhibits researchers from adequately studying this invaluable problem. To tackle this, we have meticulously constructed the GLOBESUMM dataset by first collecting a wealth of multilingual news reports and restructuring them into event-centric format. Additionally, we introduce the method of protocol-guided prompting for high-quality and cost-effective reference annotation. In MCMS, we also highlight the challenge of conflicts between news reports, in addition to the issues of redundancies and omissions, further enhancing the complexity of GLOBESUMM. Through extensive experimental analysis, we validate the quality of our dataset and elucidate the inherent challenges of the task. We firmly believe that GLOBESUMM, given its challenging nature, will greatly contribute to the multilingual communities and the evaluation of LLMs.

Figures (7)

• Figure 1: The protocol, we formulated for MCMS task, includes definitions of all potential error types, along with corresponding real-life examples for each type and approximate resolution strategies. (English version in Figure \ref{['fig:error_types_en']})
• Figure 2: Overview of our silver-quality summary annotation methodology. The method consists of key information split, cross-lingual prompting and protocol-guided prompting.
• Figure 3: $P^*$, $R^*$ and $F_1^*$ scores of GPT-4 in [where]. All values are calculated in micro-averaging.
• Figure 4: Average error rates of LLMs for each type of conflict as a proportion of the total errors.
• Figure 5: The average proportion of content in summaries generated by LLMs that is entailed in different source documents across 26 languages.