An investigation on opinion mining in cross domain and sentiment analysis

Abstract

Opinion mining is a developing field to track the opinion of people newlinewhich they expose in form of their feedback on different area Recently newlineseveral research works are designed for sentiment analysis based on newlineclassification and ranking methods However the amount of time required for newlineperforming classification was not reduced Then Sentiment Sensitivity newlineThesaurus SST was presented to solve the feature mismatch difficulty in newlinecross domain sentiment classification among source and target domains newlineHowever achieving higher accuracy and detecting distributional similarities newlineof words was not effectual Initially Hidden Markov Continual Progression Cosine Similar HM-CPCS method is introduced in order to attain efficient pre-processing newlineaccuracy Proposed HM-CPCS method employs the POS tagger according to newlinethe Hidden Markov with reviews gathered from dissimilar domains Besides newlinethe subsequent and antecedent tags estimate the transition and word newlineoccurrence probability among the tags for enhancing feature extraction newlineaccuracy with better efficient Then the stemmer model is employed to newlineexpand the reviews during training and test times for evaluating the transition newlineprobabilities of the review words according to generated observations Lastly newlinethe similarity function is utilized for measuring the relatedness of cross newlinedomain algorithm based on cosine factor by applying HM-CPCS method. newline newline