Grammatical Relations in Chinese: GB-Ground Extraction and Data-Driven Parsing

Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics, pages 446–456, Baltimore, Maryland, USA, June 23-25 2014. c⃝2014 Association for Computational Linguistics Grammatical Relations in Chinese: GB-Ground Extraction and Data-Driven Parsing Weiwei Sun, Yantao Du, Xin Kou, Shuoyang Ding, Xiaojun Wan∗ Institute of Computer Science and Technology, Peking University The MOE Key Laboratory of Computational Linguistics, Peking University {ws,duyantao,kouxin,wanxiaojun}@pku.edu.cn, dsy100@gmail.com Abstract This paper is concerned with building linguistic re- sources and statistical parsers for deep grammatical relation (GR) analysis of Chinese texts. A set of linguistic rules is defined to explore implicit phrase structural information and thus build high-quality GR annotations that are represented as general di- rected dependency graphs. The reliability of this linguistically-motivated GR extraction procedure is highlighted by manual evaluation. Based on the converted corpus, we study transition-based, data- driven models for GR parsing. We present a novel transition system which suits GR graphs better than existing systems. The key idea is to introduce a new type of transition that reorders top k elements in the memory module. Evaluation gauges how successful GR parsing for Chinese can be by applying data- driven models. 1 Introduction Grammatical relations (GRs) represent functional relationships between language units in a sen- tence. They are exemplified in traditional gram- mars by the notions of subject, direct/indirect object, etc. GRs have assumed an important role in linguistic theorizing, within a variety of approaches ranging from generative grammar to functional theories. For example, several com- putational grammar formalisms, such as Lexi- cal Function Grammar (LFG; Bresnan and Ka- plan, 1982; Dalrymple, 2001) and Head-driven Phrase Structure Grammar (HPSG; Pollard and Sag, 1994) encode grammatical functions directly. In particular, GRs can be viewed as the depen- dency backbone of an LFG analysis that provide general linguistic insights, and have great potential advantages for NLP applications, (Kaplan et al., 2004; Briscoe and Carroll, 2006; Clark and Cur- ran, 2007a; Miyao et al., 2007). ∗Email correspondence. In this paper, we address the question of an- alyzing Chinese sentences with deep GRs. To acquire high-quality GR corpus, we propose a linguistically-motivated algorithm to translate a Government and Binding (GB; Chomsky, 1981; Carnie, 2007) grounded phrase structure treebank, i.e. Chinese Treebank (CTB; Xue et al., 2005) to a deep dependency bank where GRs are ex- plicitly represented. Different from popular shal- low dependency parsing that focus on tree-shaped structures, our GR annotations are represented as general directed graphs that express not only lo- cal but also various long-distance dependencies, such as coordinations, control/raising construc- tions, topicalization, relative clauses and many other complicated linguistic phenomena that goes beyond shallow syntax (see Fig. 1 for example.). Manual evaluation highlights the reliability of our linguistically-motivated GR extraction algorithm: The overall dependency-based precision and recall are 99.17 and 98.87. The automatically-converted corpus would be of use for a wide variety of NLP tasks. Recent years have seen the introduction of a number of treebank-guided statistical parsers ca- pable of generating considerably accurate parses for Chinese. With the high-quality GR resource …

Content truncated to 500 words