Concept mining for indexing medical literature

This article addresses the task of mining concepts from biomedical literature to index and search through a documents base. This research takes place within the Telemakus project, which has for goal to support and facilitate the knowledge discovery process by providing retrieval, visual, and interaction tools to mine and map research findings from research literature in the field of aging. A concept mining component automating research findings extraction such as the one presented here, would permit Telemakus to be efficiently applied to other domains. The main strategy that has been followed in this project has been to mine front the legends of the documents the research findings as relationships between concepts from the medical literature. The concept mining proceeds through stages of syntactic analysis, semantic analysis, relationships building, and ranking. Evaluation results are presented at the end and show that the system learns concepts and relationships between them with good recall, and that these concepts can be used for indexing the documents. Future improvements of the system are also presented. (C) 2006 Elsevier Ltd. All rights reserved.