Generating Sports News from Live Commentary: A Chinese Dataset for Sports Game Summarization

Proceedings of the 1st Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 10th International Joint Conference on Natural Language Processing, pages 609–615 December 4 - 7, 2020. c⃝2020 Association for Computational Linguistics 609 Generating Sports News from Live Commentary: A Chinese Dataset for Sports Game Summarization Kuan-Hao Huang1 Chen Li2 Kai-Wei Chang1 1University of California, Los Angeles 2Tencent AI Lab 1{khhuang, kwchang}@cs.ucla.edu 2lichen.lc18@gmail.com Abstract Sports game summarization focuses on gen- erating news articles from live commentaries. Unlike traditional summarization tasks, the source documents and the target summaries for sports game summarization tasks are written in quite different writing styles. In addition, live commentaries usually contain many named en- tities, which makes summarizing sports games precisely very challenging. To deeply study this task, we present SPORTSSUM1, a Chinese sports game summarization dataset which con- tains 5,428 soccer games of live commentaries and the corresponding news articles. Addi- tionally, we propose a two-step summarization model consisting of a selector and a rewriter for SPORTSSUM. To evaluate the correctness of generated sports summaries, we design two novel score metrics: name matching score and event matching score. Experimental results show that our model performs better than other summarization baselines on ROUGE scores as well as the two designed scores. 1 Introduction There are a large number of sports games playing every day. Apparently, manually writing sports news articles to summarize every game is labor- intensive and infeasible. How to automatically gen- erate sports summaries, therefore, becomes a popu- lar and demanding task. Recently, generating news from live commentaries has gradually attracted at- tention in the academic community (Zhang et al., 2016; Yao et al., 2017). At the same time, several trials have been done in the industry such as sports news from Toutiao’s Xiaoming Bot2, Sohu Ruibao3 and AI football news4. 1The dataset is available at https://github.com/ ej0cl6/SportsSum 2http://www.nbd.com.cn/columns/803 3https://mp.sohu.com/profile?xpt= c29odW1wMzZpdDlzQHNvaHUuY29t 4https://www.51zhanbao.com Live Commentary Time Scores Commentary Sentence 66’ 0-0 多特蒙德球员格策拼抢犯规,对手获得控球权. Dortmund’s player G¨otze fouled, and the opponent got the possession of the ball. 66’ 0-0 施魏因斯泰格为拜仁慕尼黑赢得一个任意球. Schweinsteiger got a free kick for Bayern Munich. 67’ 1-0 进球啦！！！拜仁慕尼黑球员克罗斯大禁区外 左脚射门,球从右下角飞进球门,球进了!助攻的 是穆勒. 拜仁慕尼黑1-0 多特蒙德. Goal!!! Bayern Munich’s player Kroos shot with his left foot from the outside of the penalty area. The ball flew into the goal through the lower right corner. The ball went in! Muller gave the assist. Bayern Munich 1-0 Dortmund. 71’ 1-0 拜仁慕尼黑球员里贝里大禁区左侧尝试右脚射 门,可惜皮球高出球门.给他传球的是拉姆. Bayern Munich’s player Ribery tried to shoot with his right foot from the penalty area’s left side, but the ball was higher than the crossbar. Lahm passed the ball to him. Sports News Ariticle 开场3分钟，克罗斯左侧任意球被顶到后点，里贝里禁区边缘抽射 偏出近门柱。第8分钟，穆勒右路与曼朱基奇打出踢墙配合，在门 前12米处推射被苏博蒂奇铲出底线。第13分钟，里贝里右路塞球， 克罗斯在门前27米处抽射偏出近门柱。(...) In the 3rd minutes, Kroos’s free kick on the left was tipped to the back, and Ribery’s shot from the penalty area missed. In the 8th minute, Muller and Mandzukic had teamwork, and Muller’s shot from the 12 meters ahead the goal line was touched out by Suboti´c. In the 13th minute, Ribery passed the ball from the right, and Kroos’s shot near the 27 …

Content truncated to 500 words