ETimeline: An Extensive Timeline Generation Dataset based on Large Language Model
Xiaochen Liu, Yanan Zhang
TL;DR
ETimeline addresses the shortage of large-scale, diverse timeline datasets by introducing a bilingual dataset with $600$ timelines and $13{,}878$ event nodes across $28$ domains, built from a pool of over $120{,}000$ articles. The authors propose a two-phase data-construction pipeline driven by a $7$B-parameter LLM, with four sub-tasks (topic generation, refining, mounting, deduplication) and knowledge distillation from GPT-$4$, yielding a strong, industry-relevant baseline. The dataset supports tasks such as topic generation and event relationship modeling, and includes the original news pool to facilitate further research. Ethical considerations emphasize TOS compliance, toxicity reduction, and access to data via titles and URLs to maintain accessibility while protecting content.
Abstract
Timeline generation is of great significance for a comprehensive understanding of the development of events over time. Its goal is to organize news chronologically, which helps to identify patterns and trends that may be obscured when viewing news in isolation, making it easier to track the development of stories and understand the interrelationships between key events. Timelines are now common in various commercial products, but academic research in this area is notably scarce. Additionally, the current datasets are in need of refinement for enhanced utility and expanded coverage. In this paper, we propose ETimeline, which encompasses over $13,000$ news articles, spanning $600$ bilingual timelines across $28$ news domains. Specifically, we gather a candidate pool of more than $120,000$ news articles and employ the large language model (LLM) Pipeline to improve performance, ultimately yielding the ETimeline. The data analysis underscores the appeal of ETimeline. Additionally, we also provide the news pool data for further research and analysis. This work contributes to the advancement of timeline generation research and supports a wide range of tasks, including topic generation and event relationships. We believe that this dataset will serve as a catalyst for innovative research and bridge the gap between academia and industry in understanding the practical application of technology services. The dataset is available at https://zenodo.org/records/11392212