On the Organization and Retrieval of Health QA Records for Community-based Health Services

The urgent needs for online health information resulted in the establishment of many community-based health services such as HealthTap1 and HaoDF2. They offer knowledge in the form of question answer (QA) pairs, where questions are issued by grassroot healthseekers, and answers are provided by doctors. These services have some intrinsic properties. First, they are crowdsourcing data that are continually growing at fast pace, and it is thus not practical to organize them manually. Second, they are unstructured and unlabeled in terms of topics, which greatly hinder their retrieval and browsing. Third, health seekers and doctors with diverse backgrounds present the same concepts in colloquial style, which leads to wide vocabulary gap. Together, these pose big challenges for data access and navigation. Recent efforts [Akbari et al., 2016] indicate that organizing the community-contributed data into a hierarchical structure may enhance coarse-grained browsing and fined-grained search.