Exploiting the semantic graph for the representation and retrieval of medical documents.

Objective: The objective of this study was to propose a graph-based semantic search approach by addressing the inherent complexity and ambiguity of medical terminology in queries and clinical text for enhanced medical information retrieval.

Methods: The supportive use of a medical domain ontology exploits the light-weight semantics discovered from queries and documents for enhanced document ranking. First, the implicit information regarding concepts and the relations between them is discovered in the documents and queries and is used to evaluate the relevance of the query-document; then, the semantic linkages between concepts distributed in target documents and reference documents are built and used to score the document's popularity; finally, the above two evaluations are integrated to produce the final ranking list for document ranking.

Results: Empirical experiments are conducted on two different datasets. The results demonstrate that the proposed graph-based approach significantly outperforms the baselines. For example, the average performance improvement on two datasets of the best variant of GSRM compared to the best baseline achieve 7.2% and 7.9% in terms of P@20 and NDCG@20, respectively, which illustrates the effectiveness of the proposed approach.