Category: Novosti

On 28-30 September 2023 the ReLDI Centre and several members of the ReLDI network will be in Rijeka for the CLARC 2023 conference, organised by the Center for Language Research of the Faculty of Humanities and Social Sciences at the University of Rijeka. The conference topic is “Language and Language Data”, to be discussed in ReLDI light through plenary talks by Maja Miličević Petrović and Nikola Ljubešić, a panel on the recently completed UPSKILLS project, and a roundtable on “Linguistics and Large Language Models”. In thanking the organisers for having us, we look forward to seeing many familiar faces and meeting new ones!

The ventures of ReLDI are now described in a book chapter, available in open access!

Within the newly published book CLARIN. The Infrastructure for Language Resources, which marks CLARIN’s 10th anniversary as a European Research Infrastructure Consortium, Nikola Ljubešić, Tomaž Erjavec, Maja Miličević Petrović and Tanja Samardžić tell the story of how ReLDI came about, how it help(ed|s) the development of technologies for South Slavic languages, and how its initial contacts with colleagues from Slovenia grew into a long-standing collaboration with CLARIN.SI and the CLASSLA knowledge centre. Together we are stronger!

On 15 and 16 September 2022 the ReLDI Centre and the ReLDI network had several representatives at the JTDH conference in Ljubljana. We are especially pround of two young researchers from Belgrade – Natalija Tomić and Ružica Farmakovski – who presented work born from ReLDI events and supervised by Tanja Samardžić and Nikola Ljubešić. Check out the proceedings for more details!

An important moment in the conference was also a ReLDI-CLASSLA get-together. The rather spontaneous meeting started as a get-to-meet-everyone-in-a-circle (first picture), and developed into an NLP 101 talk by Boshko Koloski (second picture) and a rather wide and interesting discussion. We hope that everyone’s impressions are as positive as ours!

ReLDI centre for linguistic data will participate, as an associate partner, in a project dedicated to language technologies and digital text processing (Cr. “Jezične tehnologije i digitalna obrada teksta”). This interesting project has been approved within the call “UNIRI CLASS – Open Personalised Education”, and its objective is to introduce a new minor through a collaboration of multiple UNIRI component units. The project coordinator is Benedikt Perak.

The ReLDI centre is continuing the collaboration with CLARIN.SI’s Knowledge centre for South Slavic languages – CLASSLA. In fact, the recent success story published by CLASSLA is dedicated to the importance of collaborations and synergies, and the term “ReLDI effect” is used to describe synergistic effects between different projects. You can read the full story at https://www.clarin.si/info/k-centre/success-stories/.

In addition, the ReLDI centre has recently been involved in the development of the first open speech-to-text system for Croatian, coordinated by CLASSLA. The system is available at https://huggingface.co/classla/wav2vec2-xls-r-parlaspeech-hr, where it is possible to try out some examples, but also upload or record one’s own speech. The work continues, and ParlaSpeech-HR is planned to be published in early 2022. All this is the results of joint efforts by Nikola Ljubešić, Ivo-Pavao Jazbec, Vuk Batanović, Lenka Bajčetić, Danijel Korzinek and Peter Rupnik, but would not have been possible without a wider collaboration around the ParlaMint project, with Darja Fišer, Tomaž Erjavec, Maciej Ogrodniczuk and Petya Osenova.

The materials from Regional variation in gender marking: a hands-on tutorial on extracting data from corpora are now available for download from https://github.com/clarinsi/workshop_reg_mark. These materials provide an introduction to the process of using corpora to study a linguistic (and not only linguistic) problem, with information on:

• how to find (comparable) South Slavic corpora in the CLARIN.SI repository
• how to explore corpora through the noSketchEngine and KonText concordancers
• how to study gender marking looking at frequencies of feminine and masculine nouns describing occupations, and at the distribution of feminine and masculine forms of different verbs
• how to draw conclusions about gender bias in society based on corpus results

The materials were prepared by Mirjana Starović and Tanja Samardžić as part of the online workshop held on 6 and 7 November 2021, organised by the University of Zurich – URPP “Language and Space”, the CLARIN knowledge centre for South Slavic languages – CLASSLA and the ReLDI centre. The programme also included a keynote talk by Yves Scherrer from the University of Helsinki, Darja Fišer’s presentation of opportunities for student presentations at the JTDH Language Technologies and Digital Humanities Conference, and an Interactive workshop on regional variation in text led by Sara Košutar, Larissa Schmidt and Leyla Feiner.

The workshop saw the participation of around 30 students and colleagues divided between GatherTown and Zoom, with lively and fun interactive sessions, and some surprising findings. A follow-up mentoring session for students took place on 16 December 2021.

For CLASSLA accounts of the workshop, see here:

Workshops

Workshop on regional markedness in text

The University of Zurich – URPP “Language and Space”, the CLARIN knowledge centre for South Slavic languages – CLASSLA and the ReLDI centre are organising an online workshop dedicated to regional markedness in text. The workshop will take place on Zoom, 6-7 November 2021. The details, including the programme and the registration link, can be found here. Students are particularly encouraged to apply!

On 26 February 2021, ReLDI will participate in the conference Internationalisms in Slavic as a window into the architecture of grammar – InterSlavic 2020/2021, organised by the University of Graz. The abstract and the presentation can be found here, and in case you wish to join us and the rest of the conference, the registration instructions are available here.

The CLASSLA knowledge centre has recently published a new state-of-the-art BERT-like model covering Croatian and Serbian. The model is available via HuggingFace, and a version fine-tuned to the task of named entity recognition has been published on the same platform too (see here). The creator of the model is Nikola Ljubešić.

ReLDI centre has recently collaborated with the Research Centre of the Slovenian Academy of Sciences and Arts on the creation of corpora of language-related news articles and news comments in Serbian, Croatian and Slovene. The work was coordinated by Vuk Batanović, and the corpora can be downloaded here.