Ensemble Semantics for Large-scale Unsupervised Relation Extraction

Proceedings of the 2012 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning, pages 1027–1037, Jeju Island, Korea, 12–14 July 2012. c⃝2012 Association for Computational Linguistics Ensemble Semantics for Large-scale Unsupervised Relation Extraction Bonan Min1* Shuming Shi2 Ralph Grishman1 Chin-Yew Lin2 1New York University 2Microsoft Research Asia New York, NY, USA Beijing, China {min,grishman}@cs.nyu.edu {shumings,cyl}@microsoft.com Abstract Discovering significant types of relations from the web is challenging because of its open nature. Unsupervised algorithms are developed to extract relations from a cor- pus without knowing the relations in ad- vance, but most of them rely on tagging arguments of predefined types. Recently, a new algorithm was proposed to jointly extract relations and their argument se- mantic classes, taking a set of relation in- stances extracted by an open IE algorithm as input. However, it cannot handle poly- semy of relation phrases and fails to group many similar (“synonymous”) rela- tion instances because of the sparseness of features. In this paper, we present a novel unsupervised algorithm that provides a more general treatment of the polysemy and synonymy problems. The algorithm incorporates various knowledge sources which we will show to be very effective for unsupervised extraction. Moreover, it explicitly disambiguates polysemous rela- tion phrases and groups synonymous ones. While maintaining approximately the same precision, the algorithm achieves significant improvement on recall com- pared to the previous method. It is also very efficient. Experiments on a real- world dataset show that it can handle 14.7 million relation instances and extract a very large set of relations from the web. 1 Introduction Relation extraction aims at discovering semantic relations between entities. It is an important task that has many applications in answering factoid questions, building knowledge bases and improv- ing search engine relevance. The web has become a massive potential source of such relations. How- ever, its open nature brings an open-ended set of relation types. To extract these relations, a system should not assume a fixed set of relation types, nor rely on a fixed set of relation argument types. The past decade has seen some promising solu- tions, unsupervised relation extraction (URE) algo- rithms that extract relations from a corpus without knowing the relations in advance. However, most algorithms (Hasegawa et al., 2004, Shinyama and Sekine, 2006, Chen et. al, 2005) rely on tagging predefined types of entities as relation arguments, and thus are not well-suited for the open domain. Recently, Kok and Domingos (2008) proposed Semantic Network Extractor (SNE), which gener- ates argument semantic classes and sets of synon- ymous relation phrases at the same time, thus avoiding the requirement of tagging relation argu- ments of predefined types. However, SNE has 2 limitations: 1) Following previous URE algo- rithms, it only uses features from the set of input relation instances for clustering. Empirically we found that it fails to group many relevant relation instances. These features, such as the surface forms of arguments and lexical sequences in between, are very sparse in practice. In contrast, there …

Content truncated to 500 words