Twitter Topic Summarization by Ranking Tweets using Social Influence and Content Quality

Proceedings of COLING 2012: Technical Papers, pages 763–780, COLING 2012, Mumbai, December 2012. Twitter Topic Summarization by Ranking Tweets Using Social Influence and Content Quality DUAN YaJuan1∗ CHEN ZhuMin2 W EI FuRu3 ZHOU Ming3 Heung − Yeung SHUM 4 (1) University of Science and Technology of China, No. 96, Jinzhai Road, Hefei, Anhui, P.R.China (2) Shandong University, No. 27, Shanda South Road, Jinan, Shandong, P.R.China (3) Microsoft Research Asia, No. 5, Danling Street, Haidian District, Beijing, P.R.China (4) Microsoft Corporation, Redmond, USA ❞②❥❅♠❛✐❧✳✉st❝✳❡❞✉✳❝♥✱ ❝❤❡♥③❤✉♠✐♥❅s❞✉✳❡❞✉✳❝♥ ④❢✉✇❡✐✱ ♠✐♥❣③❤♦✉✱ ❤s❤✉♠⑥❅♠✐❝r♦s♦❢t✳❝♦♠ ABSTRACT In this paper, we propose a time-line based framework for topic summarization in Twitter. We summarize topics by sub-topics along time line to fully capture rapid topic evolution in Twitter. Specifically, we rank and select salient and diversified tweets as a summary of each sub-topic. We have observed that ranking tweets is significantly different from ranking sentences in traditional extractive document summarization. We model and formulate the tweet ranking in a unified mutual reinforcement graph, where the social influence of users and the content quality of tweets are taken into consideration simultaneously in a mutually reinforcing manner. Extensive experiments are conducted on 3.9 million tweets. The results show that the proposed approach outperforms previous approaches by 14% improvement on average ROUGE-1. Moreover, we show how the content quality of tweets and the social influence of users effectively improve the performance of measuring the salience of tweets. TITLE AND ABSTRACT IN ANOTHER LANGUAGE (CHINESE) ú ú úŽ Ž Ž( ( (7 7 7> > >� � �q q qÍ Í Í› › ›Œ Œ Œ‡ ‡ ‡, , ,( ( (Ï Ï Ï„ „ „¨ ¨ ¨y y yÝ Ý Ý˜ ˜ ˜X X X• • • ,‡Ðú†�*úŽöôt„¨yݘê¨X•F¶�ݘ�göôz•��Pݘ�� Pݘ-9n‡,„Í•¦Œ�7'ù¨y‡,ÛL’•Œ½Öå��X•��ì•(ø ’ž:�þ!‹�ö�Q‡,…¹�\�>�qÍ›Œ‡,(ÏÛL’•�žŒÓœh ��1�,‡Ðú„!‹(úÆ!‹�ROUGE-1sGÐ؆14%�2�\�>�qÍ›Œ‡ ,(Ï�H09Û†‡,Í•¦„¦Ï� KEYWORDS: Topic summarization, Twitter, Timeline summarization, Content quality, Social influence. KEYWORDS IN CHINESE: ݘX•�¨y�öôt�„X•�‡,(Ï�(7>�qÍ › . ∗ This work was done when the first author was an intern at Microsoft Research Asia. 763 1 Introduction Recently, Twitter 1 has become one of the most popular social networking sites. It enables people to freely post short messages (called tweets) up to 140 characters. Twitter has rapidly gained worldwide popularity, with over 140 million active users generating over 340 million tweets daily in March 2012 2. The rapid proliferation of Twitter posts presents a big obstacle for efficient information acquisition. It is impossible for a user to get an overview of important topics on Twitter by reading all tweets everyday. In addition, because of information redundancy and the informal writing style, it is time consuming to find useful information about a topic from a huge number of tweets. The tremendous volume of tweets suggests summarization as the key to facilitating the requirements of topic exploration, navigation, and search from hundreds of thousands of tweets. Specifically, a summary that provides representative information of topics with no redundancy and well-written sentences would be preferred. In this paper, we focus on the problem of topic summarization in Twitter, which aims to provide a short and compact …

Content truncated to 500 words