You are here
Development of a Test Collection of Manually Extracted Semantic Relationships in Health Consumer Texts
Semantic relationships within knowledge bases are the links that connect concepts to one another. They are often used, for example, within information retrieval applications for search term expansion. The overall goal of this project was to manually identify the semantic relationships within health consumer question and physician-provided answer texts. We created a collection of manually identified semantic relationships for purposes of evaluating automated extraction methods. We identified a total of 509 semantic relationship instances within twelve consumer-oriented question-answer pairs (avg. of 275 words per pair). Coding of the semantic relationships was based on a set of revised relations derived from the Unified Medical Language System (UMLS) Semantic Network.