Context-based Retrieval and Visualization of Medical Information
Medical information retrieval on the World-Wide Web has been a main way for healthcare information consumers to explore, compare, and learn medical information. To promote the utility (and hence the real value) of medical information, automatic retrieval and visualization of medical information are essential. They are often triggered by a passage of interest description (e.g. a prescription, research issue, research finding, and disease symptom) from the user. The passage is often short and in natural language form. It may serve as a good starting point from which the user navigates through the space of relevant passages. The navigation is a key to the discovery and cross-validation of medical knowledge, which are particularly important for the healthcare domain. In this project, we explore how and to what extent context of discussion (COD) of passages may be recognized to support the retrieval and visualization of relevant passages. Fundamental techniques will be developed, including (1) context-based classifier building, (2) context-based query generation and text retrieval, and (3) identification and visualization of relevant passages. Theoretical and empirical evaluation will be conducted, and a system that integrates the techniques will be implemented to automatically interpret the given passage, generate suitable queries, retrieve relevant passages, and visualize the relevant passages for the user to explore. The contributions are of both theoretical and practical significance to promoting the utility and real value of medical information to human life.
Keywords: Medical information retrieval, recognition of context of discussion, text classification, query generation, text retrieval, passage retrieval, passage visualization.