Enhancing Universal Dependencies for Korean

Noh, Youngbin; jiyoon, Han; Oh, Tae Hwan; Kim, Hansaem

doi:10.18653/v1/w18-6013

Cited by 5 publications

(4 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our new guidelines follow the UDv2 guidelines. UPOS and DEPREL are revised to reflect Korean morphological features and flexible word-order aspects with reference to Korean UD studies such as Park et al (2018), Oh (2019). In UPOS, ADJ, NOUN, and VERB are revised extensively.…”

Section: Discussionmentioning

confidence: 99%

“…This effort resulted in the establishment of standard annotation guidelines of Korean dependencies, giving rise to various related efforts that focused on the establishment of Korean UD guidelines that better represent the unique Korean linguistic features. These studies include Park et al (2018) who focused on the mapping between the UD part-of-speech (POS) tags and the POS tags in the Sejong Treebank, and and Oh (2019) who provided in-depth discussions of applicability and relevance of UD's dependency relation to Korean.…”

Section: Korean Ud Corporamentioning

confidence: 99%

“…In re-examining PTK-UD v2018's dependency relations, we consulted two existing dependency annotation guidelines for Korean: and Oh (2019). They offer a thorough analysis on applicability of the universal dependency relation labels to Korean, and further identify a list of dependency relations such as iobj, xcomp, expl, and cop (among others) as not suited for capturing characteristics of Korean grammar.…”

Section: Deprel Revisionmentioning

confidence: 99%

See 2 more Smart Citations

Proceedings of the 16th International Conference on Parsing Technologies and the IWPT 2020 Shared Task on Parsing into Enhanced Universal Dependencies

2020

View full text Add to dashboard Cite

Welcome to the 16th International Conference on Parsing Technologies (IWPT 2020), which this year (for the first time since 2007) is co-located with the Annual Meeting of the Association for Computational Linguistics (ACL). The IWPT meeting series, hosted by the ACL Special Interest Group in Natural Language Parsing (SIGPARSE), has been held biennualy since its inaugual meeting in 1989 in Pittsburgh, PA (USA).For 2020, the SIGPARSE steering group decided to try out something new, co-location with the main ACL meeting in the form of a reduced one-day IWPT programme. The main motivation for this move was to reduce fragmentation (and travel) and to increase IWPT visibility in the 'mainstream' ACL community (we already know that at least one of these goals was attained). At the same time, IWPT launches its own series of parsing shared tasks this year, which strengthens the experimental and applied perspective on parsing technologies in the conference programme.The IWPT 2020 shared task focuses on the parsing of Enhanced Universal Dependencies (EUD) over 17 languages. This is the first time that graph-based representations of syntactic structures are evaluated on such a large scale, and we believe it will pave way for research on richer models and representations. The task attracted system submissions from ten teams from around the world and, thus, establishes a highly relevant point of comparison for this line of syntactic analysis. We are very grateful to everyone who contributed to this shared task, starting with the data providers who worked hard to meet our deadline. Thanks to the participant teams who worked tirelessly in a short time period to provide such a set of great and interesting systems! Owing to the COVID-19 pandemic this year, the meeting will regrettably be held entirely virtual, where for IWPT we have adopted a mostly-asynchronous format: Accepted papers (of three different types: long, short, and shared task) will be presented through pre-recorded talks, which become available online for individual viewing before the actual conference day. On the original date of the conference, July 9, there will be a three-hour live session, scheduled so that the timing should be convenient (all things considered) for participants around the world: 14:00-17:00 UTC, which translates, for example, into a starting time at 7:00 in the morning at the US West Coast and wrapping up at 1:00 in the morning in Melbourne, Australia. The live sessions will be devoted exclusively to questions and answers, organized into five thematic sessions. Authors of papers associated with each session will be available to answer questions and disucss their work (possibly also among themselves).There has been (and to some degree still is) much uncertainty about the format of ACL and IWPT this year, and in a sense we were positively surprised to receive a number of submissions comparable to recent IWPT instances. Out of 24 regular paper submissions, the programme committee accepted 14 for presentation at the conference. The IWPT 2020 progr...

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Korean Ud Corporamentioning

confidence: 99%

Section: Deprel Revisionmentioning

confidence: 99%

See 1 more Smart Citation

Proceedings of the 16th International Conference on Parsing Technologies and the IWPT 2020 Shared Task on Parsing into Enhanced Universal Dependencies

2020

View full text Add to dashboard Cite

show abstract

“…Cf. the analysis byNoh et al (2018).15 All resources are available at https://github. com/OfirArviv/translated_tacred.16 https://translate.yandex.com/ 17 https://papago.naver.com/…”

mentioning

confidence: 99%

On the Relation between Syntactic Divergence and Zero-Shot Performance

Arviv¹,

Nikolaev²,

Karidi³

et al. 2021

Preprint

View full text Add to dashboard Cite

We explore the link between the extent to which syntactic relations are preserved in translation and the ease of correctly constructing a parse tree in a zero-shot setting. While previous work suggests such a relation, it tends to focus on the macro level and not on the level of individual edges-a gap we aim to address. As a test case, we take the transfer of Universal Dependencies (UD) parsing from English to a diverse set of languages and conduct two sets of experiments. In one, we analyze zero-shot performance based on the extent to which English source edges are preserved in translation. In another, we apply three linguistically motivated transformations to UD, creating more cross-lingually stable versions of it, and assess their zero-shot parsability. In order to compare parsing performance across different schemes, we perform extrinsic evaluation on the downstream task of cross-lingual relation extraction (RE) using a subset of a popular English RE benchmark translated to Russian and Korean. 1 In both sets of experiments, our results suggest a strong relation between cross-lingual stability and zero-shot parsing performance.

show abstract