Microblog Entity Linking by Leveraging Extra Posts

Proceedings of the 2013 Conference on Empirical Methods in Natural Language Processing, pages 863–868, Seattle, Washington, USA, 18-21 October 2013. c⃝2013 Association for Computational Linguistics Microblog Entity Linking by Leveraging Extra Posts Yuhang Guo, Bing Qin∗, Ting Liu , Sheng Li Research Center for Social Computing and Information Retrieval School of Computer Science and Technology Harbin Institute of Technology, China {yhguo, bqin∗, tliu, sli}@ir.hit.edu.cn Abstract Linking name mentions in microblog posts to a knowledge base, namely microblog entity linking, is useful for text mining tasks on mi- croblog. Entity linking in long text has been well studied in previous works. However few work has focused on short text such as mi- croblog post. Microblog posts are short and noisy. Previous method can extract few fea- tures from the post context. In this paper we propose to use extra posts for the microblog entity linking task. Experimental results show that our proposed method significantly im- proves the linking accuracy over traditional methods by 8.3% and 7.5% respectively. 1 Introduction Microblogging services (e.g. Twitter) are attracting millions of users to share and exchange their ideas and opinions. Millions of new microblog posts are generated on such open broadcasting platforms ev- ery day 1. Microblog provides a fruitful and instant channel of global information publication and acqui- sition. A necessary step for the information acquisition on microblog is to identify which entities a post is about. Such identification can be challenging be- cause the entity mention may be ambiguous. Let’s begin with a real post from Twitter. (1) No excuse for floods tax, says Abbott URL ∗Corresponding author 1See http://blog.twitter.com/2011/06/ 200-million-tweets- per-day.html. This post is about an Australia political lead- er, Tony Abbot, and his opinion on flood tax policy. To understand that this post mentions Tony Abbot is not trivial because the name Ab- s.bot can refer to many people and organization- In the Wikipedia page of Abbott, there list- s more than 20 Abbotts, such as baseball player Jim Abbott, actor Bud Abbott and company Abbott Laboratories, etc.. Given a knowledge base (KB) (e.g. Wikipedia), entity linking is the task to identify the referent KB entity of a target name mention in plain text. Most current entity linking techniques are designed for long text such as news/blog articles (Mihalcea and Csomai, 2007; Cucerzan, 2007; Milne and Witten, 2008; Han and Sun, 2011; Zhang et al., 2011; Shen et al., 2012; Kulkarni et al., 2009; Ratinov et al., 2011). Entity linking for microblog posts has not been well studied. Comparing with news/blog articles, microblog posts are: short each post contains no more than 140 charac- ters; fresh the new entity-related content may have not been included in the knowledge base; informal acronyms and spoken language writing style are common. Due to these properties, few feature can be ex- tracted from a post. Without enough features, pre- vious entity linking methods may fail. In order to overcome the feature sparseness, we turn to another property of microblog: 863 redundancy For each day, over …

Content truncated to 500 words