Ekstraksi informasi leksikal bahasa Kerinci dari kamus dwibahasa untuk klasifikasi Part-of-Speech otomatis

Abstrak

This study aims to develop initial linguistic resources for the Kerinci language by extracting lexical information from an Indonesian–Kerinci bilingual dictionary. The dictionary data, originally in digital document format (.docx), was converted into a structured dataset suitable for computational processing. Part-of-Speech (POS) classification was performed automatically using a rule-based approach, referring to the Indonesian equivalents that had been POS-tagged using reference lexicons. The results show that verbs and adjectives dominate the dictionary entries, differing from the common trend in bilingual dictionaries which are usually dominated by nouns. This approach proves effective in low-resource conditions and serves as a preliminary step toward the development of Kerinci NLP. The resulting POS-classified dataset has great potential for training POS taggers, supporting linguistic documentation, and contributing to language preservation through technology