Utilization of Summarization Algorithms for a Better Understanding of Clustered Medical Documents

Medical documents contain rich information about the diseases, medication, symptoms and precautions. Extraction of useful information from large volumes of medical documents that are generated by electronic health record systems is a complex task as they are unstructured or semi-structured. Various partitional and agglomerative clustering techniques are applied for grouping the medical documents into meaningful clusters [4]. Multi-document summarization techniques which are recent development in the field of Natural Language Processing are applied to condense the huge data present in the clustered medical documents to generate a single summary which conveys the key meaning. The summarization techniques can be broadly classified into two types [2]. They are: Extractive Summarization techniques and Abstractive Summarization techniques. Extractive Summarization techniques try to retrieve the most important sentences from the given document. Abstractive Summarization techniques try to generate summary with new sentences which are not present in the document. Extractive summarization techniques using Statistical Approaches are applied on the clustered medical documents. Medical summaries help the patients for a better and prior understanding of the disease and they can get a brief idea before consulting a physician. The generated summaries are evaluated using ROUGE (Recall Oriented Understudy of Gisting Evaluation) evaluation technique.