English-Lithuanian Language Resources for the Cybersecurity Domain
Abstract
This presentation was delivered as a part of the colloquium “Data Matters: Data Management and Storage for Humanities with CLARIN-LT” at the 7th International Conference of Applied Linguistics “Languages and People” (25–27 September 2025 at Vilnius University, Lithuania) organised by the Lithuanian Association of Applied Linguistics. The colloquium sought to contribute to the ongoing discourse on data management and storage in the humanities by highlighting the importance of data management planning, presenting a certified data storage repository that provides international access to Lithuanian data, and showcasing bilingual language resources for a specialized domain. This presentation introduces five cybersecurity language resources deposited in the CLARIN-LT repository between 2022 and 2024. All of these resources were developed by the same team of researchers from Mykolas Romeris University and Vytautas Magnus University in Lithuania, beginning with the joint research project DVITAS (Bilingual Automatic Terminology Extraction) and continuing through subsequent initiatives. For more details, see the conference website: https://litaka.lt/en/7-conference/ ; https://sitti.vdu.lt/dvitas/en
Full text
EnglishLithuanian Language Resources for the Cybersecurity Domain Sigita Rackevičienė Mykolas Romeris University
2 Five bilingual language resources for the cybersecurity domain are deposited in CLARIN-LT repository, two more are being prepared for the depositing.
3
4 Language resources for the Cybersecurity (CS) domain are being created as an integral part of the project and its post-project activities on CS terminology.
5 Why Cybersecurity? Relevance – global digital connectivity & cloud-based services Security – challenges of securing sensitive data Dynamics – rapid evolution of the domain Standardisation – need to harmonise Lithuanian terminology
6 Application of Language Resources: Technology •Development of deep learning systems for automatic term extraction Terminology Management •Creation of a bilingual termbase Terminology Research •Analysis of conceptual, linguistic & pragmatic aspects •Analysis of sociolinguistic aspects
7 Parallel and comparable corpora
8 Corpora in CLARIN-LT repository English-Lithuanian comparable and parallel corpora are openly available in the CLARIN-LT repository: •Comparable corpus: plain texts and morphologically annotated texts •Parallel corpora: aligned TMX files (the 1st version and the updated 2nd version)
9 Corpora on Sketch Engine platform
16 Macrostructure of the termbase
17 Microstructure of a termbase (structure of a terminological entry)
18 Microstructure of the termbase
19 Statistics of the current terminological data in the termbase: 233 concepts 514 LT terms, 414 EN terms The number of entries is expected to increase considerably by the end of 2025, with a total of 500 entries planned.
20 Ongoing Work on the Termbase Existing Entries: •Refinement of thematic grouping of concepts •Refinement of definitions according to ISO standards •Enrichment of examples based on updated corpora •Creation of structured format for source data •Update of frequency labels of existing terms New Entries: •Extension of concept index based on updated corpora •Collection and structuring of necessary data for new concept entries Additional Work: •Cross-check with Terminų bankas and IATE datasets
21 Survey dataset
22 The survey dataset is available in the CLARIN-LT repository in TSV (TabSeparated Values) format, which can be opened in spreadsheet programs or read by programming languages. All responses are anonymous.
23 The aim of the survey The aim of the survey was to investigate terminology preferences of different user groups, including: -the choice or proposal of the most suitable terms for given concepts, -the rationale behind these selections.
24 Concepts selected for the terminology survey Concept 1 ‘cyberattack’ Concept 6 ‘phishing’ Concept 2 ‘spam’ Concept 7 ‘botnet’ Concept 3 ‘denial-of-service attack’ Concept 8 ‘hacker’ Concept 4 ‘man-in-the-middle attack’ Concept 9 ‘honeypot method’ Concept 5 ‘brute force attack’ Concept 10 ‘zero-day vulnerability’
25 ‘Zero-day vulnerability’ Sąvokos apibrėžtis: ką tik išsiaiškinta, bet dar nespėta pašalinti sistemos ar įrenginio saugumo spraga (šaltinis: IATE). Angliškas terminas: zero-day vulnerability Concept definition: vulnerability in a system or device that has been disclosed but is not yet patched (source: IATE) English term: zero-day vulnerability Which term, in your opinion, is most suitable to desingate this concept? Why have you chosen this term / proposed your variant? (you can select several reasons)
32 Future Plans for Depositing to CLARIN-LT: •CS parallel LT–EN corpus (Lithuanian national documents and their English translations) •Version 2 of the CS comparable corpus •Manually annotated CS training corpus (for training deep learning systems - automatic term extraction) •Version 2 of the CS termbase (TBX) Further Development: Application of LLOD technologies for linking the CS language resources with other web-based resources.
33 Main research papers on the presented language resources (including figures on slides 11, 17 and 28): 1) Utka, A., Rackevičienė, S, Mockienė, L., Rokas, A., Laurinaitis, M., Bielinskienė, A. 2022. Building of Parallel and Comparable Cybersecurity Corpora for Bilingual Terminology Extraction. Selected Papers from the CLARIN Annual Conference 2021. Linköping Electronic Conference Proceedings 189, 126–138. 2) Rackevičienė, S., Utka, A., Bielinskienė, A., Mockienė, L. 2023. Lithuanian-English Cybersecurity Termbase: Principles of Data Collection and Structuring. Rasprave Instituta za hrvatski jezik, 49 (2), 439-461. 3) Rackevičienė, S., Utka, A. 2024. Preferences of Lithuanian Cybersecurity Synonymous Terms in Different User Groups. Studies about Languages / Kalbų studijos 44, 107-122. For more information on the project, visit the project website https://sitti.vdu.lt/dvitas/en