IndoSum: A New Benchmark Dataset for Indonesian Text Summarization

TL;DR

IndoSum is a large, publicly available dataset for Indonesian text summarization, enabling better research and benchmarking in low-resource language NLP tasks.

Contribution

This paper introduces IndoSum, a significantly larger Indonesian summarization dataset, and provides baseline evaluations for extractive summarization methods.

Findings

The dataset is nearly 200 times larger than previous Indonesian datasets.

Baseline extractive methods show promising results on IndoSum.

The dataset and code are openly accessible for future research.

Abstract

Automatic text summarization is generally considered as a challenging task in the NLP community. One of the challenges is the publicly available and large dataset that is relatively rare and difficult to construct. The problem is even worse for low-resource languages such as Indonesian. In this paper, we present IndoSum, a new benchmark dataset for Indonesian text summarization. The dataset consists of news articles and manually constructed summaries. Notably, the dataset is almost 200x larger than the previous Indonesian summarization dataset of the same domain. We evaluated various extractive summarization approaches and obtained encouraging results which demonstrate the usefulness of the dataset and provide baselines for future research. The code and the dataset are available online under permissive licenses.