Wikidata, a rapidly expanding knowledge graph (KG), owes its growth to two primary factors. First, Wikidata allows open access and editing by anyone. Second, it offers a multilingual feature that enables data entities to be accessed in various languages worldwide. However, the issue of incomplete information across multiple languages remains a significant challenge. For instance, the description of the entity “bada reuteuk” (ID: Q100606305) is currently available only in Indonesian as “a traditional food in Indonesia,” but it lacks descriptions in other languages. Consequently, these data are not accessible or recognizable in languages other than Indonesian. The system incorporates two primary features: language profiling and data translation. Language profiling, implemented using SPARQL queries via the Wikidata API, provides an overview of the multilingual status of Wikidata entities. For data translation, the system utilized the Translated Labs library, chosen for its open access, cost-free availability, and high-quality translation outputs. The translated results are subsequently saved into Wikidata. System evaluation involved five respondents from the Wikidata community, using a black-box testing approach. Results demonstrated that MULTI-WD’s core functionalities—including category selection, data statistics display, translation, and data updates—achieved 100% operational success. Furthermore, the tool enhanced data translation efficiency by up to 300% compared to manual translation directly through the Wikidata interface.
Copyrights © 2024