Learning Fine-grained Relations from Chinese User Generated Categories

Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 2577–2587 Copenhagen, Denmark, September 7–11, 2017. c⃝2017 Association for Computational Linguistics Learning Fine-grained Relations from Chinese User Generated Categories Chengyu Wang, Yan Fan, Xiaofeng He∗, Aoying Zhou Shanghai Key Laboratory of Trustworthy Computing, School of Computer Science and Software Engineering, East China Normal University {chywang2013,eileen940531}@gmail.com {xfhe,ayzhou}@sei.ecnu.edu.cn Abstract User generated categories (UGCs) are short texts that reflect how people describe and organize entities, expressing rich se- mantic relations implicitly. While most methods on UGC relation extraction are based on pattern matching in English cir- cumstances, learning relations from Chi- nese UGCs poses different challenges due to the flexibility of expressions. In this pa- per, we present a weakly supervised learn- ing framework to harvest relations from Chinese UGCs. We identify is-a relations via word embedding based projection and inference, extract non-taxonomic relations and their category patterns by graph min- ing. We conduct experiments on Chinese Wikipedia and achieve high accuracy, out- performing state-of-the-art methods. 1 Introduction UGCs are descriptive phrases related to entities, frequently appearing in online encyclopedias and vertical websites. These texts are concise and in- formative, reflecting the way people organize and characterize entities (Xu et al., 2016a). UGCs (especially Wikipedia categories) are im- portant sources for knowledge harvesting. Previ- ous approaches (Flati et al., 2014; Ponzetto and Strube, 2007; Ponzetto and Navigli, 2009) focus on inferring is-a relations between entities and UGCs for taxonomy construction. A few others extract multiple types of relations from Wikipedia categories (Nastase and Strube, 2008; Suchanek et al., 2007). These methods are mostly designed for English language by employing language- specific patterns or linguistic rules. ∗Corresponding author. For Chinese, harvesting semantic relations from texts poses different challenges. There is no dis- tinction between singular and plural forms and no word spaces in Chinese. Word orders can be ar- ranged in multiple ways with very flexible expres- sions. As illustrated in Qiu and Zhang (2014); Chen et al. (2014), the research of relation extrac- tion from Chinese texts makes less significant pro- cess than the research for English. Although sev- eral approaches are proposed to construct Chinese taxonomies from Wikipedia categories (Li et al., 2015; Wang et al., 2014), extracting fine-grained and multi-typed relations from UGCs still needs further study. This is because there exist very few high-quality lexical patterns for relation iden- tification in Chinese UGCs (in contrast to Nas- tase and Strube (2008); Suchanek et al. (2007)). Hence this problem is similar to “open relation ex- traction” (Etzioni et al., 2011) from Chinese short texts, without pre-defined relation types. ��·���-��Tim Berners-Lee �������Winner of Turing Award ��������1955 ������ ��������History of the Internet ����Londoner ��·���-� Tim Berners-Lee ��� Turing Award ����� 1955 ��� Londoner win-prize born-in is-a ������ Winner of Turing Award is-a Figure 1: An illustrative example with respect to “Tim Berners-Lee” in Chinese Wikipedia. In this paper, we propose a weakly super- vised learning framework to mine fine-grained and multiple-typed relations from Chinese UGCs. A simple example is illustrated …

Content truncated to 500 words