Medline document classification model

Medline is online repository of medical literature As the natural language documents are unstructured, finding relevant features are very complex problem, and non linear distribution of documents increases the problem to find the decision surface that separates the data. We proposed a prototype based classifier to resolve these problems We empirically tested the model with Medline database and bench mark dataset Reuters 21875. As Medline data contain more complex unstructured data, it need better tools to extract information from the Medline and to categorize them. With our proposed algorithm we achieved nearly 90% of accuracy to classify the documents.