Supplement to "Mind the Language Gap in Digital Humanities: LLM-Aided Translation of SKOS Thesauri"

Information This is the data supplement to "Mind the Language Gap in Digital Humanities: LLM-Aided Translation of SKOS Thesauri" which was accepted at the Computational Humanities Research Conference (CHR 2025). It includes all thesauri that were translated for the evaluation of WOKIE, a tool for automated translation of SKOS thesauri. Folder Structure - There are five different folders that contain thesauri of the following topics: - Archaeology - Cultural time periods - Interlink of Digital Humanities, computer science and IT - Historic book and library studies - Latin Additionally, there are results of ontology machting in the `om` folder. File naming convention (not applicable to the om folder) - <thesaurus>.rdf is the original, unaltered thesaurus, e.g. `defc.rdf` - Some thesauri have been altered, this is appended to the thesaurus name: - `_reduced` if some terms not being part of the selected topic have been removed. - `_converted` if there was a conversion step from another serialization to RDF/XML involved. - `_corrected_skosified` if invalid SKOS was corrected using a Python library called `skosify`to make usage possible. - To evaluate WOKIE, one language was removed from the thesauri. This is noted as `_node` if German was removed ("no German"), or `_nola` if Latin was removed. - All different test cases are in put into folders accordingly, e.g. `defc_node` for DEFC with German removed. - Within these folders, there are numerous files: - All different translated SKOS thesauri (`*.rdf`) with a descriptive filename including primary and secondary translation services that were used and a timestamp. - Result files belonging to the translated SKOS thesauri in `.csv` and `.log` - Both files contain metrics for the comparison of the original thesaurus (before language removal) and the back-translated thesaurus. - The `.log` file additionally includes information about code parameters. om folder The `om` folder contains all the test cases used for ontology matching evaluation. It was performed using the MELT test kit provided by the OAEI (ontology alignment evaluation initiative). The main results are in `testCasePerformanceCube.xlsx`. Similarity calculation folder To reproduce the results or for performing your own evaluation with different thesauri, go to the `similarity-calculation`folder and follow the instruction in the README within this folder. Licenses DEFC [CC BY 4.0; Creators: Seta Štuhec, Anja Masur, Peter Andorfer, Ksenia Zaytseva, Edeltraud Aspöck]PACTOLS (adapted) [ODbL v1.0; Creators: Groupe PACTOLS/FRANTIQ]iDAI.world (adapted) [CC BY 4.0; Creators: Annika Kirscheneder, Camilla Colombi, Elenore Pape, Gabriele Rasbach, Henriette Senst, Lena Vitt, Matthias Block, Nina Dworschak, Reinhard Förtsch, Sabine Thänert]Iron-Age-Danube [CC BY 4.0; Creator: Seta Štuhec]PARTHENOS (adapted) [CC BY 4.0; Creators: PARTHENOS project]OeAI (adapted) [CC BY 4.0; Creator: Micheline Welte]DHA Taxonomy (adapted) [CC BY 4.0; Creators: ACDH-OEAW Team]UNESCO (adapted) [CC BY-SA 3.0; Creators: UNESCO]TaDiRAH (adapted) [CC0; Creators: Luise Borek, Canan Hastik, Vera Khramova, Jonathan Geiger]