Abstract: We investigate text representation methods for bridging the gap between professional and consumer health vocabularies. By representing medical concepts as embedding vectors, we evaluate the similarity between a professional medical concept and its consumer-friendly counterpart using a ranking-based approach. On two sets of professional-consumer concept pairs, concept embedding methods can rank a professional medical concept the 14th (7th) closest to its consumer-friendly counterpart and vice versa among a large collection of medical concepts.
Learning Objective 1: A text embedding approach to mining professional-consumer medical concept pairs.
Learning Objective 2: Leveraging large-scale text corpora in medical domain for medical vocabulary construction.
Yue Wang (Presenter)
University of Michigan
Jian Tang, University of Michigan
V.G.Vinod Vydiswaran, University of Michigan
Kai Zheng, University of California, Irvine
Hua Xu, University of Texas Health Science Center
Qiaozhu Mei, University of Michigan