The Hmong Medical Corpus: a biomedical corpus for a minority language

Abstract Biomedical communication is an area that increasingly benefits from natural language processing (NLP) work. named entity recognition (NER) in particular provides a foundation for advanced NLP applications, such as automated medical question-answering and translation services. However, while large body of biomedical documents are available array languages, most work NER remains English, with the remainder official national or regional languages. Minority languages so far remain underexplored area. The Hmong language, minority sizable populations several countries without status anywhere, represents exceptional challenge effective contexts. Taking advantage number government-produced information Hmong, we have developed first entity-annotated corpus resource-poor language. Medical Corpus contains 100,535 tokens 4554 entities (NEs) three UMLS semantic types: diseases/syndromes, signs/symptoms, parts/organs/organ components. Furthermore, subset annotated word position parts speech, representing gold-standard dataset publicly Hmong. methodology presented readily reproducible approach creation NE-annotated corpora other