Automatic labelling of important terms and phrases from medical discussions

Online remedy finder web services have become quite popular in recent years. In these services, the user posts a text description of his health problem and the expert provides medications and related suggestions. Lengthy text description is the major challenge of these services as common web users often put a long text to describe his sufferings and the experts do not have sufficient time to read the long text. In this paper, we propose a system that highlights the important phrases or terms from such long questions or diagnosis related discussions. The proposed system uses a hybrid strategy where a set of statistical measures, syntactical and semantic information are used. To evaluate the system, we create a test data where all the important phrases are manually labelled. In our experiments, we found that the system achieves 83.38% recall.