Building a Corpus of Medical Paraphrases in Romanian
Résumé
Understanding medical language is a challenge for laypeople due to complex scientific formula and technical terms used in this domain. These specific terms are difficult to understand and require paraphrasing (rewriting, simplifying) for laypeople. In this preliminary study, we aim to constitute a first medical corpus, annotated with medical paraphrases in Romanian. This resource is useful for the simplification of monolexical ("hipoacuzia"; hypoacusis) and multiword medical terms in Romanian ("carcinom endometrial"; endometrial carcinoma). Our method consists of automatically searching medical terms and specific markers that help to paraphrase them. We present our methodology to collect this corpus and for the annotation process. We also present quantitative data and first experiences on qualitative annotation of the identified paraphrases.