Bag of Words Involving Selection of Term Sets Using Apriori Algorithm in Text Classification

Text mining area enables people to extract the relevant information from the given text. This approach is advantageous, because it allows the users to retrieve the useful information and helps to avoid unambiguity in the text. Text Mining tasks are classified as text classification, text clustering and summarization of documents. Phrases are important in the field of text mining and information retrieval. Phrases identification, classification of phrases constitute of a major importance because, phrases are constructed by means of term sets. Term sets or item sets constitute a set of two terms usually bigram. The occurrence of a term in the document can be found by means of providing binary weights. Document representation can be done with Bag of Words (BOW) model. The main motivation of this paper is that term sets are constructed by computing the frequency of each term in the corresponding document and weighting for a term is also provided. Terms sets involving adjacent pair and nonadjacent pair are taken into consideration and this is used for the classifying positive documents and negative documents. Association rule mining is used for the construction of term sets. News Group dataset was taken which consists of twenty thousand messages is one which is widely used. In this paper, first preprocessing was done by means of stop word removal and stemming. Finally, association rules are formed by using apriori algorithm and term sets are formed.