Multi-instance Multi-label Learning for Relation Extraction

Proceedings of the 2012 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning, pages 455–465, Jeju Island, Korea, 12–14 July 2012. c⃝2012 Association for Computational Linguistics Multi-instance Multi-label Learning for Relation Extraction Mihai Surdeanu†, Julie Tibshirani†, Ramesh Nallapati⋆, Christopher D. Manning† † Stanford University, Stanford, CA 94305 {mihais,jtibs,manning}@stanford.edu ⋆ Artificial Intelligence Center, SRI International nallapat@ai.sri.com Abstract Distant supervision for relation extraction (RE) – gathering training data by aligning a database of facts with text – is an efficient ap- proach to scale RE to thousands of different relations. However, this introduces a challeng- ing learning scenario where the relation ex- pressed by a pair of entities found in a sen- tence is unknown. For example, a sentence containing Balzac and France may express BornIn or Died, an unknown relation, or no re- lation at all. Because of this, traditional super- vised learning, which assumes that each ex- ample is explicitly mapped to a label, is not appropriate. We propose a novel approach to multi-instance multi-label learning for RE, which jointly models all the instances of a pair of entities in text and all their labels using a graphical model with latent variables. Our model performs competitively on two difficult domains. 1 Introduction Information extraction (IE), defined as the task of extracting structured information (e.g., events, bi- nary relations, etc.) from free text, has received re- newed interest in the “big data” era, when petabytes of natural-language text containing thousands of dif- ferent structure types are readily available. How- ever, traditional supervised methods are unlikely to scale in this context, as training data is either lim- ited or nonexistent for most of these structures. One of the most promising approaches to IE that ad- dresses this limitation is distant supervision, which generates training data automatically by aligning a DB = � BornIn(Barack Obama, United States) EmployedBy(Barack Obama, United States) � Sentence Latent Label Barack Obama is the 44th and current President of the United States. EmployedBy Obama was born in the United States just as he has always said. BornIn United States President Barack Obama meets with Chinese Vice President Xi Jinping today. EmployedBy Obama ran for the United States Senate in 2004. – Figure 1: Training sentences generated through distant supervision for a database containing two facts. database of facts with text (Craven and Kumlien, 1999; Bunescu and Mooney, 2007). In this paper we focus on distant supervision for relation extraction (RE), a subproblem of IE that ad- dresses the extraction of labeled relations between two named entities. Figure 1 shows a simple exam- ple for a RE domain with two labels. Distant super- vision introduces two modeling challenges, which we highlight in the table. The first challenge is that some training examples obtained through this heuristic are not valid, e.g., the last sentence in Fig- ure 1 is not a correct example for any of the known labels for the tuple. The percentage of such false positives can be quite high. For example, Riedel …

Content truncated to 500 words