Constructing Web-Accessible Semantic Role Labels and Frames for Japanese as Additions to the NPCMJ Parsed Corpus

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 3153–3161 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 3153 Constructing Web-Accessible Semantic Role Labels and Frames for Japanese as Additions to the NPCMJ Parsed Corpus Koichi Takeuchi∗, Alastair Butler⋆, Iku Nagasaki†, Takuya Okamura∗, Prashant Pardeshi‡ ∗Okayama University, ⋆Hirosaki University, †Nagoya University, ‡National Institute of Japanese Language and Linguistics {takeuc-k@, t-okamura@s}.okayama-u.ac.jp, ajb129@hirosaki.ac.jp, inagasaki@free.fr, prashant@ninjal.ac.jp Abstract As part of constructing the NINJAL Parsed Corpus of Modern Japanese (NPCMJ), a web-accessible language resource, we are adding frame information for predicates, together with two types of semantic role labels that mark the contributions of arguments. One role type consists of numbered semantic roles, like in PropBank, to capture relations between arguments in different syntactic patterns. The other role type consists of semantic roles with conventional names. Both role types are compatible with hierarchical frames that belong to related predicates. Adding semantic role and frame information to the NPCMJ will support a web environment where language learners and linguists can search examples of Japanese for syntactic and semantic features. The annotation will also provide a language resource for NLP researchers making semantic parsing models (e.g., for AMR parsing) following machine learning approaches. In this paper, we describe how the two types of semantic role labels are defined under the frame based approach, i.e., both types can be consistently applied when linked to corresponding frames. Then we show special cases of syntactic patterns and the current status of the annotation work. Keywords: semantic roles, predicate frames, sentence level meaning, PropBank, thesaurus 1. Introduction Semantic role labeled data is important to capture predicate-argument sentence level meaning for NLP re- searchers, linguists, and language learners. For English, various kinds of annotated corpora with semantic roles and frames are provided (e.g., PropBank (Bonial et al., 2010) and FrameNet (Baker et al., 1998)) and they are widely used. The numbered semantic roles (e.g., Arg0, Arg1, Arg2) proposed in PropBank are effective for adapting to various systems of semantic roles. Thus, the Abstract Meaning Representation (AMR) (Banarescu et al., 2013) approach, which is based on PropBank numbered seman- tic roles and frames, has attracted a lot of attention from NLP researchers as a model for describing sentence-level semantics. AMR parsing models are studied with deep neu- ral network models, e.g., (Zhang et al., 2019) using AMR resources for English1. In terms of Japanese language resources, several cor- pora containing annotated tags related to predicate- argument information have been proposed (Kyoto (Kawahara et al., 2002), NAIST (Iida et al., 2007), EDR (EDR, 1995), GDA (Hashida, 2005), Japanese FrameNet (Ohara et al., 2006), BCCWJ-PT (Takeuchi et al., 2015)). However, numbered semantic roles have not been an- notated for Japanese texts. Also, none of the annotated corpora that are reported in the literature are web accessible because of licensing restrictions2. 1LDC2017T10 and LDC2014T12. 2Japanese FrameNet is constructed on the Balanced Corpus of Contemporary Written Japanese (BCCWJ) (Maekawa et al., 2014), from which it inherits release …

Content truncated to 500 words