Examining the State-of-the-Art in News Timeline Summarization

Demian Gholipour Ghalandari,Georgiana Ifrim

doi:10.18653/v1/2020.acl-main.122

Abstract

Previous work on automatic news timeline summarization (TLS) leaves an unclear picture about how this task can generally be approached and how well it is currently solved. This is mostly due to the focus on individual subtasks, such as date selection and date summarization, and to the previous lack of appropriate evaluation metrics for the full TLS task. In this paper, we compare different TLS strategies using appropriate evaluation frameworks, and propose a simple and effective combination of methods that improves over the stateof-the-art on all tested benchmarks. For a more robust evaluation, we also present a new TLS dataset, which is larger and spans longer time periods than previous datasets.

Highlights

Timelines of news events can be useful to condense long-ranging news topics and can help us understand how current major events follow from prior events
While timeline summarization (TLS) has been studied before, most works treat it as a combination of two individual subtasks, 1) date selection and 2) date summarization, and only focus on one of these at a time (Tran et al, 2013a,b, 2015b)
For a more robust evaluation, we present a new dataset for TLS, which is significantly larger than previous datasets in terms of the number of individual topics and time span

Summary

Introduction

Timelines of news events can be useful to condense long-ranging news topics and can help us understand how current major events follow from prior events. While TLS has been studied before, most works treat it as a combination of two individual subtasks, 1) date selection and 2) date summarization, and only focus on one of these at a time (Tran et al, 2013a,b, 2015b). These subtasks are almost never evaluated in combination, which leaves an unclear picture of how well TLS is being solved in general. Previously used evaluation metrics for the date selection and timeline summarization tasks are not appropriate since they do not consider the temporal alignment in the evaluation. There were no established experimental settings and appropriate metrics for the full TLS task (Martschat and Markert, 2017, 2018)

Methods

Discussion

Conclusion