IRCMS at SemEval-2018 Task 7 : Evaluating a basic CNN Method and Traditional Pipeline Method for Relation Classification

Proceedings of the 12th International Workshop on Semantic Evaluation (SemEval-2018), pages 811–815 New Orleans, Louisiana, June 5–6, 2018. ©2018 Association for Computational Linguistics IRCMS at SemEval-2018 Task 7 : Evaluating a basic CNN Method and Traditional Pipeline Method for Relation Classification Zhongbo Yin, Zhunchen Luo, Wei Luo∗, Bin Mao, Changhai Tian, Yuming Ye, Shuai Wu Information Research Center of Military Science, PLA Academy of Military Science, China zhongboyin@foxmail.com; {zhunchenluo,lwowen79}@gmail.com; {miscy210,jamestch,yuming ye}@163.com; wus1986@gmail.com Abstract This paper presents our participation for sub- task1 (1.1 and 1.2) in SemEval 2018 task 7: Semantic Relation Extraction and Classifica- tion in Scientific Papers (G´abor et al., 2018). We experimented on this task with two meth- ods: CNN method and traditional pipeline method. We use the context between two en- tities (included) as input information for both methods, which extremely reduce the noise effect. For the CNN method, we construct a simple convolution neural network to auto- matically learn features from raw texts with- out any manual processing. Moreover, we use the softmax function to classify the entity pair into a specific relation category. For the tradi- tional pipeline method, we use the Hackabout method as a representation which is described in section3.5. The CNN method’s result is much better than traditional pipeline method (49.1% vs. 42.3% and 71.1% vs. 54.6% ). 1 Introduction Scientific paper, as a major source of new tech- nology, is a common way for tracing the dynam- ics of a research domain. With lots of papers published every year, scholars can’t read all of them to extract useful aspects for our research do- main. Information extraction (IE) is a main NLP aspects for analyzing scientific papers, which in- cludes named entity recognition (NER) and rela- tion extraction (RE). Scientific papers’ informa- tion extraction is identify concepts or semantic re- lation between these concepts. This paper focuses on the relation classification between relative con- cepts in scientific paper. Relation classification is one of the most impor- tant topics for analyzing scientific papers. Most of the traditional relation classification methods ∗Corresponding author are influenced by the handcrafted features or ex- tra NLP tools to derive lexical features (Surdeanu et al., 2012; Kozareva, 2012). However, these methods are time consuming and there is a prob- lem of error propagation. Additionally, the tra- ditional semantic textual similarity measuring ap- proaches are using a large number of pairwise sim- ilarity features to represent the text. It is difficult for these features to represent the syntactic infor- mation. To address these problems, DNN methods have been proposed and made remarkable achievement (Qin et al., 2016a; Guo et al., 2016). This paper is based on the work of (Qin et al., 2016b) which uses a CNN architecture to control feature learn- ing automatically. As a result, Qin et al. (2016b) minimize the application of external toolkits and resources, which is used for part of speech (POS) or other basic pretreatment. Additionally, Zeng et al. (2014) proposes position feature to locate the entity pair, so as to highlight its …

Content truncated to 500 words