Search

This paper investigates whether adding data of typologically closer languages improves the performance of transformer-based models for three different downstream tasks, namely Part-of- Speech tagging, Named Entity Recognition, and Sentiment Analysis, compared to a monolingual and plain multilingual language model. For the presented pilot study, we performed experiments for the use case of Slovene, a low(er)-resourced language belonging to the Slavic language group. The experiments were carried out in a controlled setting, where a monolingual model for Slovene was compared to combined language models containing Slovene, trained with the same amount of Slovene data. The experimental results show that adding typologically closer languages indeed improves the performance of the Slovene language model, and even succeeds in outperforming the large multilingual XLM-RoBERTa model for NER and PoStagging. We also reveal that, contrary to intuition, distant or unrelated languages also combine admirably with Slovene, often outperforming XLM-R as well. All the bilingual models used in the experiments are publicly available.

This contribution presents version 1.5 of the Annotated Corpora for Term Extraction Research (ACTER) dataset. It includes domain-specific corpora in three languages (English, French, and Dutch) and four domains (corruption, dressage (equitation), heart failure, and wind energy). Man- ual annotations are available of terms and Named Entities for each corpus, with almost 20k unique annotations in total. Significant improvements have been made, most notably the inclu- sion of sequential annotations. Additionally, an online demo – D-Termine – has been launched for monolingual and bilingual automatic term extraction from parallel corpora, based on the dataset. ispartof: pages:1-4 ispartof: CLARIN Annual Conference Proceedings, 2022 pages:1-4 ispartof: CLARIN Annual Conference location:Prague, Czechia date:10 Oct - 12 Oct 2022 status: published

Annotated company-specific events in economic news articles. Also contains replication data for Jacobs G, et al. "Economic Event Detection in Company-Specific News Text"

Accompanying data and metadata for the preprint and submission of Van Hee, Jacobs, Emmery, et al. 2018. Automatic Detection of Cyberbullying in Social Media Text.

This contribution presents D-Terminer: an open access, online demo for monolingual and multilingual automatic term extraction from parallel corpora. The monolingual term extraction is based on a recurrent neural network, with a supervised methodology that relies on pretrained embeddings. Candidate terms can be tagged in their original context and there is no need for a large corpus, as the methodology will work even for single sentences. With the bilingual term extraction from parallel corpora, potentially equivalent candidate term pairs are extracted from translation memories and manual annotation of the results shows that good equivalents are found for most candidate terms. Accompanying the release of the demo is an updated version of the ACTER Annotated Corpora for Term Extraction Research (version 1.5). ispartof: pages:33-40 ispartof: Proceedings of the Workshop on Terminology in the 21st Century : Many Faces, Many Places, co-located with the LREC 2022 conference pages:33-40 ispartof: Workshop on Terminology in the 21st Century : Many Faces, Many Places, co-located with the LREC 2022 conference location:Marseille, France date:20 Jun - 25 Jun 2022 status: published

This paper presents the results of a replication experiment for automatic irony detection in Dutch social media text, investigating both a feature-based SVM classifier, as was done by Van Hee et al. (2017) and a transformer-based approach. In addition to building a baseline model, an important goal of this research is to explore the implementation of commonsense knowledge in the form of implicit sentiment, as we strongly believe that commonsense and connotative knowledge are essential to the identification of irony and implicit meaning in tweets. We show promising results and how the presented approach can provide a solid baseline and serve as a staging ground to build on in future experiments for irony detection in Dutch.

In this paper, we present the SentEMO platform, a tool that provides aspect-based sentiment analysis and emotion detection of unstructured text data such as reviews, emails and customer care conversations. Currently, models have been trained for five domains and one general domain and are implemented in a pipeline approach, where the output of one model serves as the input for the next. The results are presented in three interactive dashboards, allowing companies to gain more insights into what stakeholders think of their products and services. The SentEMO platform is available at https://sentemo.ugent.be/(1).

As emotion analysis in text has gained a lot of attention in the field of natural language processing, differences in emotion expression across languages could have consequences for how emotion detection models work. We evaluate the language-dependence of an mBERT-based emotion detection model by comparing language identification performance before and after fine-tuning on emotion detection, and performing (adjusted) zero-shot experiments to assess whether emotion detection models rely on language-specific information. When dealing with typologically dissimilar languages, we found evidence for the language-dependence of emotion detection.

This paper reports on a set of proof-of-concept experiments performed to evaluate and improve the alignment of monolingual embeddings for a specialised domain, viz. the medical use case of heart failure. The presented approach, which creates domain-specific dictionaries on-the-fly from cross-lingual Wikipedia links, achieves good results for cross-lingual alignment of this specialised vocabulary in three language pairs: English-Dutch, English-French, and Dutch-French. The experimental results show that the setup incorporating a smaller but dedicated domain-specific dictionary outperforms the alignment incorporating a larger but general-domain seed dictionary. A detailed error analysis reveals that many potentially useful (near-)equivalents are found beyond those present in the gold standard, and it inspires strategies for further improvements, such as lemmatisation and improved tokenisation. ispartof: Computational Linguistics in the Netherlands Journal vol:12 pages:125-140 status: published

This paper reports on experiments for cross-lingual transfer using the anchor-based approach of Schuster et al. (2019) for English and a low-resourced language, namely Hindi. For the sake of comparison, we also evaluate the approach on three very different higher-resourced languages, viz. Dutch, Russian and Chinese. Initially designed for ELMo embeddings, we analyze the approach for the more recent BERT family of transformers for a variety of tasks, both mono and cross-lingual. The results largely prove that like most other cross-lingual transfer approaches, the static anchor approach is underwhelming for the low-resource language, while performing adequately for the higher resourced ones. We attempt to provide insights into both the quality of the anchors, and the performance for low-shot cross-lingual transfer to better understand this performance gap. We make the extracted anchors and the modified train and test sets available for future research at https://github.com/pranaydeeps/Vyaapak

In this research, we present pilot experiments to distil monolingual models from a jointly trained model for 102 languages (mBERT). We demonstrate that it is possible for the target language to outperform the original model, even with a basic distillation setup. We evaluate our methodology for 6 languages with varying amounts of resources and belonging to different language families.

We describe the creation of CLARIN Belgium (CLARIN-BE) and, associated with that, the plans of the CLARIN-VL consortium within the CLARIAH-VL infrastructure for which funding was secured for the period 2021-2025.

This paper presents a pilot study to automatic linguistic preprocessing of Ancient and Byzantine Greek, and morphological analysis more specifically. To this end, a novel subword-based BERT language model was trained on the basis of a varied corpus of Modern, Ancient and Post-classical Greek texts. Consequently, the obtained BERT embeddings were incorporated to train a fine-grained Part-of-Speech tagger for Ancient and Byzantine Greek. In addition, a corpus of Greek Epigrams was manually annotated and the resulting gold standard was used to evaluate the performance of the morphological analyser on Byzantine Greek. The experimental results show very good perplexity scores (4.9) for the BERT language model and state-of-the-art performance for the fine-grained Part-of-Speech tagger for in-domain data (treebanks containing a mixture of Classical and Medieval Greek), as well as for the newly created Byzantine Greek gold standard data set. The language models and associated code are made available for use at https://github.com/pranaydeeps/Ancient-Greek-BERT

Talking about odors and flavors is difficult for most people, yet experts appear to be able to convey critical information about wines in their reviews. This seems to be a contradiction, and wine expert descriptions are frequently received with criticism. Here, we propose a method for probing the language of wine reviews, and thus offer a means to enhance current vocabularies, and as a by-product question the general assumption that wine reviews are gibberish. By means of two different quantitative analyses—support vector machines for classification and Termhood analysis—on a corpus of online wine reviews, we tested whether wine reviews are written in a consistent manner, and thus may be considered informative; and whether reviews feature domain-specific language. First, a classification paradigm was trained on wine reviews from one set of authors for which the color, grape variety, and origin of a wine were known, and subsequently tested on data from a new author. This analysis revealed that, regardless of individual differences in vocabulary preferences, color and grape variety were predicted with high accuracy. Second, using Termhood as a measure of how words are used in wine reviews in a domain-specific manner compared to other genres in English, a list of 146 wine-specific terms was uncovered. These words were compared to existing lists of wine vocabulary that are currently used to train experts. Some overlap was observed, but there were also gaps revealed in the extant lists, suggesting these lists could be improved by our automatic analysis.

ispartof: pages:85-94 ispartof: Proceedings of the 6th International Workshop on Computational Terminology (COMPUTERM 2020), co-located with LREC 2020 pages:85-94 ispartof: The 6th International Workshop on Computational Terminology (COMPUTERM 2020), co-located with LREC 2020 location:online date:16 May - 16 May 2020 status: published

This paper presents two different systems for the SemEval shared task 7 on Assessing Humor in Edited News Headlines, sub-task 1, where the aim was to estimate the intensity of humor generated in edited headlines. Our first system is a feature-based machine learning system that combines different types of information (e.g. word embeddings, string similarity, part-of-speech tags, perplexity scores, named entity recognition) in a Nu Support Vector Regressor (NuSVR). The second system is a deep learning-based approach that uses the pre-trained language model RoBERTa to learn latent features in the news headlines that are useful to predict the funniness of each headline. The latter system was also our final submission to the competition and is ranked seventh among the 49 participating teams, with a root-mean-square error (RMSE) of 0.5253.

One of the major challenges currently facing the field of argumentation mining is the lack of consensus on how to analyse argumentative user-generated texts such as online comments. The theoretical motivations underlying the annotation guidelines used to generate labelled corpora rarely include motivation for the use of a particular theoretical basis. This pilot study reports on the annotation of a corpus of 100 Dutch user comments made in response to politically-themed news articles on Facebook. The annotation covers topic and aspect labelling, stance labelling, argumentativeness detection and claim identification. Our IAA study reports substantial agreement scores for argumentativeness detection (0.76 Fleiss’ kappa) and moderate agreement for claim labelling (0.45 Fleiss’ kappa). We provide a clear justification of the theories and definitions underlying the design of our guidelines. Our analysis of the annotations signal the importance of adjusting our guidelines to include allowances for missing context information and defining the concept of argumentativeness in connection with stance. Our annotated corpus and associated guidelines are made publicly available.