STUDY OF DOMAIN CLASSIFICATION OF RANDOM OFFLINE AND ONLINE DATA

Abstract

The search engine like google provides record of results that shows a list of ranked output. The ranking does not consider the subject of the file. The results of search engine are not in a well-defined group. This may also be frustrating, as the users have to scroll through many inappropriate results. This could come up when the user is a beginner or has superficial capabilities about the domain of interest, however more as a rule it is due to the question being brief and ambiguous. One answer is to organize search results through categorization, in specific, the classification. A goal of testing is to test designed on a controlled data set, which shows that classification-bounded search could enhance the person’s search expertise in terms of the numbers of results the person would must check out earlier than pleasing his/her query. This work uses the naive bayes classifier, which is a simple and effective method for establishing classifiers. The proposed model for finding domain, related to user query based on document index matrix. The proposed implementation combine the both approach simultaneously which is term based and phrased based. Document index matrix used term, phrased based document matrix in such a manner that it is compare with training data, and put them into relatively domain. The naïve bayes algorithm used to find maximum probability occurrence from both the matrix. The output comes in the form of suggestion domains list. user easily retrieve the data with minimum time.