Knowledge-Based Document Retrieval in Medical Domain

In this paper, we present a knowledge-based search engine that can assist ordinary users to search for medical information. User prefers to receive comprehensive, relevant information from the search results. An ordinary user is usually unfamiliar with medical background. We introduce a knowledge-based search which utilizes semantic knowledge to help understanding the documents. Medical concepts are identified from medical text with the help of MMTx. The binary relations between these concepts are also extracted based on some semantic and syntactic constraints. Documents are then matched based on similarity these two kinds of knowledge related to the user query. Experiment shows the semantic expressiveness of concepts and relations improve the performance of retrieval.