DANGNT@UIT.VNU-HCM at SemEval 2019 Task 1: Graph Transformation System from Stanford Basic Dependencies to Universal Conceptual Cognitive Annotation (UCCA)

Proceedings of the 13th International Workshop on Semantic Evaluation (SemEval-2019), pages 97–101 Minneapolis, Minnesota, USA, June 6–7, 2019. ©2019 Association for Computational Linguistics 97 1 DANGNT@UIT.VNU-HCM at SemEval 2019 Task 1: Graph Trans- formation System from Stanford Basic Dependencies to Universal Conceptual Cognitive Annotation (UCCA) Dang Tuan Nguyen and Trung Tran University of Information Technology, VNU-HCM Ho Chi Minh City, Vietnam dangnt@uit.edu.vn, ttrung@nlke-group.net Abstract This paper describes the graph transfor- mation system (GT System) for SemEval 2019 Task 1: Cross-lingual Semantic Pars- ing with Universal Conceptual Cognitive Annotation (UCCA)1. The input of GT System is a pair of text and its unannotated xml, which is a layer 0 part of UCCA form. The output of GT System is the cor- responding full UCCA xml. Based on the idea of graph illustration and transfor- mation, we perform four main tasks when building GT System. At the first task, we illustrate the graph form of stanford dependencies2 of input text. We then trans- form into an intermediate graph in the sec- ond task. At the third task, we continue to transform into ouput graph form. Finally, we create the output UCCA xml. The evaluation results show that our method generates good-quality UCCA xml and has a meaningful contribution to the semantic representation sub-field in Natu- ral Language Processing. 1 Introduction In the past few years, semantic representation is receiving growing attention in NLP. Researchers have recently proposed different semantic schemes. Examples include Abstract Meaning Representation (Banarescu et al. 2013), Broad- coverage Semantic Dependencies (Oepen et al. 2014), Universal Decompositional Semantics (White et al. 2016), Parallel Meaning Bank (Ab- zianidze et al. 2016), Universal Conceptual Cog- nitive Annotation (Abend and Rappoport 2013). These advances in semantic representation, along with corresponding advances in semantic parsing, text understanding, summarization, paraphrase detection, and semantic evaluation. In SemEval 2019 Task 1: Cross-lingual Se- mantic Parsing with Universal Conceptual Cogni- tive Annotation (UCCA)1, the Committee focuses on parsing text according to the UCCA semantic annotation. UCCA (Abend and Rappoport 2013) is a cross-linguistically applicable semantic repre- sentation scheme, based on Basic Linguistic The- ory (Dixon 2010). In general, UCCA represents the semantics of linguistic utterances as directed acyclic graphs (DAGs). In one DAG, nodes and edges belong to one of several layers. There are two types of node: (i) terminal nodes express the text tokens; (ii) non-terminal nodes express se- mantic units. Edges are labelled, indicating the role of a child in the relation the parent represents. As an example, consider sentence in Example 1: “The album was recorded in Switzerland .”. Two · layers of UCCA xml of this sentence: Layer0: <root annotationID="0" passageID="503012"> <attributes /> <layer layerID="0"> <attributes /> <extra ... /> <node ID="0.1" type="Word"> <attributes ... text="The" /> <extra dep="det" ... tag="DT" /> </node> ... </layer> </root> The relations of NodeID and corresponding lexi- con: {[ID="0.1" è dep="det" è “The”] [ID="0.2" è dep="nsubj:pass" è “album”] [ID="0.3" è dep="aux:pass" è “was”] [ID="0.4" è dep="root" è “recorded”] [ID="0.5" è dep="case" è “in”] [ID="0.6" è dep="obl" è “Switzerland”] …

Content truncated to 500 words