FBK-IRST: Semantic Relation Extraction Using Cyc

Proceedings of the 5th International Workshop on Semantic Evaluation, ACL 2010, pages 214–217, Uppsala, Sweden, 15-16 July 2010. c⃝2010 Association for Computational Linguistics FBK-IRST: Semantic Relation Extraction using Cyc Kateryna Tymoshenko and Claudio Giuliano FBK-IRST I-38050, Povo (TN), Italy tymoshenko@fbk.eu, giuliano@fbk.eu Abstract We present an approach for semantic re- lation extraction between nominals that combines semantic information with shal- low syntactic processing. We propose to use the ResearchCyc knowledge base as a source of semantic information about nominals. Each source of information is represented by a specific kernel func- tion. The experiments were carried out using support vector machines as a clas- sifier. The system achieves an overall F1 of 77.62% on the “Multi-Way Classifica- tion of Semantic Relations Between Pairs of Nominals” task at SemEval-2010. 1 Introduction The SemEval-2010 Task 8 “Multi-Way Classifi- cation of Semantic Relations Between Pairs of Nominals” consists in identifying which seman- tic relation holds between two nominals in a sen- tence (Hendrickx et al., 2010). The set of rela- tions is composed of nine mutually exclusive se- mantic relations and the Other relation. Specifi- cally, the task requires to return the most informa- tive relation between the specified pair of nomi- nals e1 and e2 taking into account their order. An- notation guidelines show that semantic knowledge about e1 and e2 plays a very important role in dis- tinguishing among different relations. For exam- ple, relations Cause-Effect and Product-Producer are closely related. One of the restrictions which might help to distinguish between them is that products must be concrete physical entities, while effects must not. Recently, there has emerged a large number of freely available large-scale knowledge bases. The ground idea of our research is to use them as source of semantic information. Among such re- sources there are DBpedia,1 YAGO,2 and Open- Cyc.3 On the one hand, DBpedia and YAGO have been automatically extracted from Wikipedia. They have a good coverage of named entities, but their coverage of common nouns is poorer. They seem to be more suitable for relation extraction be- tween named entities. On the other hand, Cyc is a manually designed knowledge base, which de- scribes actions and entities both in common life and in specific domains (Lenat, 1995). Cyc has a good coverage of common nouns, making it in- teresting for our task. The full version of Cyc is freely available to the research community as Re- searchCyc.4 We approached the task using the system intro- duced by Giuliano et al. (2007) as a basis. They exploited two information sources: the whole sen- tence where the relation appears, and WordNet synonymy and hyperonymy information. In this paper, we (i) investigate usage of Cyc as a source of semantic knowledge and (ii) linguistic infor- mation, which give useful clues to semantic re- lation extraction. From Cyc, we obtain informa- tion about super-classes (in the Cyc terminology generalizations) of the classes which correspond to nominals in a sentence. The sentence itself provides linguistic information, such as local con- texts of entities, …

Content truncated to 500 words