TRANSRW at SemEval-2018 Task 12: Transforming Semantic Representations for Argument Reasoning Comprehension

Proceedings of the 12th International Workshop on Semantic Evaluation (SemEval-2018), pages 1142–1145 New Orleans, Louisiana, June 5–6, 2018. ©2018 Association for Computational Linguistics TRANSRW at SemEval-2018 Task 12: Transforming Semantic Representations for Argument Reasoning Comprehension Zhimin Chen, Wei Song∗, Lizhen Liu Information Engineering College Capital Normal University Beijing 100048, China {zmchen, wsong, liz liu7480}@cnu.edu.cn Abstract This paper describes our system in SemEval- 2018 task 12: Argument Reasoning Compre- hension. The task is to select the correct war- rant that explains reasoning of a particular ar- gument consisting of a claim and a reason. The main idea of our methods is based on the as- sumption that the semantic composition of the reason and the warrant should be close to the semantic representation of the corresponding claim. We propose two neural network mod- els. The first one considers two warrant can- didates simultaneously, while the second one processes each candidate separately and then chooses the best one. We also incorporate sentiment polarity by assuming that there are kinds of sentiment associations between the reason, the warrant and the claim. The exper- iments show that the first framework is more effective and sentiment polarity is useful. 1 Introduction Argument reasoning is a key step in the process of argumentation mining and is a very challeng- ing task in natural language processing and artifi- cial intelligence. Maccartney and Manning (2008) suggested that the key factor in the study of natu- ral language understanding is the mastery of nat- ural language reasoning. When we argue for an argument, it is necessary to reconstruct the implic- it reasoning (Newman and Marshall, 1992; Haber- nal et al., 2017) under the relevant assumption and premise to get a simple and concise explanation of the whole reasoning process. The Argument Reasoning Comprehension task is defined as following: Given an argument consisting of a claim C and a reason R, the goal is to select the correct war- rant that explains reasoning of this particular ar- ∗*corresponding author gument. There are only two options W0 and W1 given and only one answer is correct. Our solution is based on the assumption that the semantic composition of the reason and the true warrant should be close to the semantic represen- tation of the claim. We propose two frameworks. First one is dependent on the task settings that two warrant candidates are considered simultaneously to make a decision. The second one is more gener- al that the task is simplified as determine whether a warrant candidate can explain argument reason- ing. We found that the first one performed better. In addition, we attempt to incorporate sentimen- t polarity to capture the sentiment association be- tween the reason, the warrant and the claim. The experimental results demonstrate that adding sen- timent polarity can improve the performance. The final result is produced by an ensemble ap- proach that combines the outputs of multiple s- ingle models. Our system achieves an accuracy of 0.67 on development dataset and 0.57 on test dataset. 2 System …

Content truncated to 500 words