CLTS: A New Chinese Long Text Summarization Dataset

Abstract

The abstractive methods lack of creative ability is particularly a problem in automatic text summarization. The summaries generated by models are mostly extracted from the source articles. One of the main causes for this problem is the lack of dataset with abstractiveness, especially for Chinese. In order to solve this problem, we paraphrase the reference summaries in CLTS, the Chinese Long Text Summarization dataset, correct errors of factual inconsistencies, and propose the first Chinese Long Text Summarization dataset with a high level of abstractiveness, CLTS+, which contains more than 180K article-summary pairs and is available online 1 . Additionally, we introduce an intrinsic metric based on co-occurrence words to evaluate the dataset we constructed. We analyze the extraction strategies used in CLTS+ summaries against other datasets to quantify the abstractiveness and difficulty of our new data and train several baselines on CLTS+ to verify the utility of it for improving the creative ability of models.

CNewSum: A Large-scale Chinese News Summarization Dataset with Human-annotated Adequacy and Deducibility Level

“…TTNews [9] is provided for NLPCC Single Document Summarization competition 2 , including 50,000 training examples with summaries and 50,000 without summaries. CLTS [10] is a Chinese summarization dataset extracted from the news website ThePaper. It contains more than 180,000 long articles and summaries written by editors of the website.…”

“…Currently, most Chinese summarization datasets are collected from Chinese social media Weibo, which are limited to a 140-character length [7,8]. Some other datasets are scraped from news websites, such as Toutiao [9] and ThePaper [10]. However, those datasets are either small-scale or of low quality.…”

An unsupervised extractive summarization method based on multi-round computation

“…In order to evaluate our algorithm more comprehensively, we made experiments on four different datasets, CNN/DailyMail [6] , NYT [14] , TTNews [7] and CLTS [8] . Both the Chinese and English datasets are composed of relatively shorter articles datasets and relatively longer articles datasets.…”