Named Entity Recognition for Tamil Biomedical Documents

Valuable Information about tamil traditional medicines are available in various forms like books, magazines and websites. These instructions are however very large and unstructured. Our system focuses on constructing a NER identification module using SVM classifier to identify named entities and to classify them into their corresponding categories. The two main categories considered are name of disorders and name of ingredients used. The system uses features such as unigrams/bigrams, case markers, substring clues and tf-idf score to classify the entities into their classes. These named entities are stored in the NE Dictionary based on their categories.