tweeDe – A Universal Dependencies treebank for German tweets

Rehbein, Ines; Ruppenhofer, Josef; Do, Bich-Ngoc

doi:10.18653/v1/w19-7811

Cited by 10 publications

(11 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, in terms of other language's noisy treebanks, the largest one is German tweets treebank tweeDe, with more than 12,000 tweets. tweeDe has an accuracy of 80.65% UAS and 72.69% LAS [20]. PoSTWITA-UD, an Italian tweet treebank is second largest Tweet treebank with 6,700 tweets and has an accuracy of 86.95% UAS and 81.5% LAS [22].…”

Section: Resultsmentioning

confidence: 99%

“…Nonetheless, a pipeline for tokenizing, tagging, and parsing the tweets was trained by them, and ensemble and distillation models were developed for parsing accuracy improvement. [20] developed and annotated TweeDe, the first German Twitter treebank, as a new training and test suite for UD parsing. TweeDe includes more than 12,000 tokens of informal private communication, annotated for PoS, morphology and UD syntactic dependencies.…”

Section: Related Workmentioning

confidence: 99%

“…These aspects highlight the need for parsers and parsing resources to be adapted to provide adequate coverage of such type of user-generated content. In response to this need, there has been growing number of research studies for parsing such noisy content for languages like English [9], [27], Italian [23], Arabic [2], and German [20] etc. However, Urdu language which is a widely spoken language in South Asia still lacks in this area of NLP.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Universal Dependencies for Urdu Noisy Text

2021

IJATCSE

View full text Add to dashboard Cite

In this paper, the process of creating a Dependency Treebank for tweetsin Urdu,a morphologically rich and less-resourced languageis described. The 500 Urdu tweets treebank iscreated by manually annotating the treebank withlemma, POS tags, morphological and syntacticrelations using the Universal Dependencies annotation scheme, adopted to the peculiarities of Urdu social media text. annotation process is evaluated through Inter-annotator agreement for dependency relations and total agreement of 94.5% and resultant weighted Kappa = 0.876was observed. The treebank is evaluated through 10-fold cross validation using Maltparserwith various feature settings. Results show average UAS score of 74%, LAS score of 62.9% and LA score of 69.8%.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Universal Dependencies for Urdu Noisy Text

2021

IJATCSE

View full text Add to dashboard Cite

show abstract

“…In the future, in order to validate our findings, we will propose a novel and wider experimental setting, where more languages are included, e.g. testing our approach on other languages for which both ironyannotated datasets and UD resources are available, such as Arabic (Ghanem et al, 2020; or German 21 (Rehbein et al, 2019).…”

Section: Discussionmentioning

confidence: 97%

Multilingual Irony Detection with Dependency Syntax and Neural Models

Cignarella¹,

Basile²,

Sanguinetti³

et al. 2020

Proceedings of the 28th International Conference on Computational Linguistics

View full text Add to dashboard Cite

This paper presents an in-depth investigation of the effectiveness of dependency-based syntactic features on the irony detection task in a multilingual perspective (English, Spanish, French and Italian). It focuses on the contribution from syntactic knowledge, exploiting linguistic resources where syntax is annotated according to the Universal Dependencies scheme. Three distinct experimental settings are provided. In the first, a variety of syntactic dependency-based features combined with classical machine learning classifiers are explored. In the second scenario, two well-known types of word embeddings are trained on parsed data and tested against gold standard datasets. In the third setting, dependency-based syntactic features are combined into the Multilingual BERT architecture. The results suggest that fine-grained dependency-based syntactic information is informative for the detection of irony.

show abstract

“…We use treebanks annotated with Universal Dependencies V2.7 (Nivre et al, 2020). For German, we use GSD, which is based on news, reviews, and Wikipedia pages, and tweeDe (Rehbein et al, 2019) as the Twitter treebank. For Italian, we use ISDT and Par-TUT, which consist of legal, news, and Wikipedia texts, plus TWITTIRÒ (Cignarella et al, 2019) and PoSTWITA (Sanguinetti et al, 2018) as the Twitter treebanks.…”

Section: Treebanksmentioning

confidence: 99%

Bidirectional Domain Adaptation Using Weighted Multi-Task Learning

Dakota¹,

Sayyed²,

Kübler³

2021

Proceedings of the 17th International Conference on Parsing Technologies and the IWPT 2021 Shared Task on Parsing Into Enhanced

View full text Add to dashboard Cite

Domain adaption in syntactic parsing is still a significant challenge. We address the issue of data imbalance between the in-domain and out-of-domain treebank typically used for the problem. We define domain adaptation as a Multi-task learning (MTL) problem, which allows us to train two parsers, one for each domain. Our results show that the MTL approach is beneficial for the smaller treebank. For the larger treebank, we need to use loss weighting in order to avoid a decrease in performance below the single task. In order to determine to what degree the data imbalance between two domains and the domain differences affect results, we also carry out an experiment with two imbalanced in-domain treebanks and show that loss weighting also improves performance in an in-domain setting. Given loss weighting in MTL, we can improve results for both parsers.

show abstract

tweeDe – A Universal Dependencies treebank for German tweets

Cited by 10 publications

References 9 publications

Universal Dependencies for Urdu Noisy Text

Universal Dependencies for Urdu Noisy Text

Multilingual Irony Detection with Dependency Syntax and Neural Models

Bidirectional Domain Adaptation Using Weighted Multi-Task Learning

Contact Info

Product

Resources

About