TECHNOLOGY SE-K-NN CLASSIFICATION ALGORITHM FOR SEMANTIC INFORMATION RETRIEVAL

Due to the continuous growth of Information retrieval system, text classification is importantly needed to find the category of new information without doing the indexing process again from preliminary. Literature presents different algorithms for classification. Among variety of algorithms, k-NN classification is simple and mostly applied benchmark algorithm for classification. In k-NN classification, similarity matching is an important process which has used different measures. But in most of the measures, semantic way of finding the similarity is completely missing so the way of including semantic keywords based on important words of the documents to the matching process will further improve the matching accuracy. By taking this as motivation, SE measure is developed for the proposed SE-K-NN classification algorithm. In the first step of the proposed algorithm, documents are pre-processed to suit for feature extraction phase and features are computed to build feature database for k-NN algorithm in the second step. In the third step, the devised measure is used for text classification using k-NN algorithm. The experimentation is done with textual database and performance is proved that the proposed algorithm reached of about 75% accuracy as compared with existing algorithm reaches the value of 73%.