PAWS: A Multi-lingual Parallel Treebank with Anaphoric Relations

Abstract

We present PAWS, a multi-lingual parallel treebank with coreference annotation. It consists of English texts from the Wall Street Journal translated into Czech, Russian and Polish. In addition, the texts are syntactically parsed and word-aligned. PAWS is based on PCEDT 2.0 and continues the tradition of multilingual treebanks with coreference annotation. The paper focuses on the coreference annotation in PAWS and its language-specific differences. PAWS offers linguistic material that can be further leveraged in cross-lingual studies, especially on coreference.

Coreference Resolution: Toward End-to-End and Cross-Lingual Systems

“…In the same direction, the CORBON 2017 shared task [36] presented a large parallel corpora with silver-standard coreference annotations, aimed at the development of projection-based approaches to coreference resolution. More recently, Nedoluzhko et al [37] presented a multi-lingual parallel treebank with gold-standard coreference annotations, consisting of English text translated into Czech, Russian, and Polish.…”

Automatyczne wykrywanie nominalnych zależności referencyjnych w polskich tekstach współczesnych

“…Eksperyment motywowany chęcią porównania opisu relacji referencyjnych w różnych językach metodami korpusowymi przeprowadziliśmy wraz z Michalem Novákiem i Anną Nedoluzhko z Uniwersytetu Karola w Pradze, tworząc wielojęzyczny korpus równoległy tekstów angielskich, czeskich, polskich i rosyjskich, anotowanych relacją koreferencji zgodnie ze wspólną metodologią (Nedoluzhko i in. 2018).…”

“…Chapter 7 presents several experiments investigating reference in a broader context of discourse relations. First was the comparison of the description of Prague Discourse Treebank-compatible reference relations in Czech, Russian and Polish with the newly implemented Parallel Annotated Wall Street Journal corpus (Nedoluzhko et al 2018). The results show variation of referential properties in different languages, both in frequency of the use of referential groups and in types of reference.…”

Cross-Lingual Coreference: The Case of Bulgarian and English

“…In their next project, (Nedoluzhko et al, 2018) further investigate the cross-lingual coreference with the PAWS parallel treebank with texts in four languages -English, Czech, Russian and Polish -by annotating and analysing not only noun, but also verb phrases.…”

“…Lots of parallel corpora used for cross-lingual coreference resolution consist of news articles (Nedoluzhko et al, 2018), (Novak, 2018) and some of them contain more than one type of texts (Grishina and Stede, 2015), (Grishina and Stede, 2017) .…”