Abstract: Opinion mining has gained increasing attention and shown great practical value in recent years. Extracting opinion words and targets is a main task in opinion mining. For the purpose of customer and business perspective, the task of scanning these reviews manually is computational burden. Hence, to process reviews automatically and summarizing them in suitable form is more efficient. The distinguished problem of producing opinion summary addresses is how to determine the mood, and opinion expressed in the review with respect to a numerical feature value. This paper proposes a novel approach with a hybrid algorithm which combines Expectation Maximation (EM) algorithm. It focused on the main task of opinion mining called as opinion summarization. The extraction of product feature, technical feature value and opinion are critical for opinion summarization as they affect the performance significantly. The proposed approach consists of a software system in which mining of product feature, technical feature value and opinion is performed. The main motto of this software system is to recognize the technical feature value depending on review, which the reviews are summarized. This software is helpful for humans to understand the technical values expressed in the reviews. It represents relations between opinion words and targets, which is employed to measure the confidence of each candidate from opinion words and targets datasets. The words or targets with high confidence are kept in their respective datasets and the rest are removed as false results which are used to refine extraction rules. K-nearest neighbor classifier - used for classify the extracted data’s in an opinion mining. Experimental results Shows the effectiveness of proposed method and finally, candidates with higher confidence are extracted as opinion targets or opinion words.

Keywords: Opinion Mining, Opinion Targets Extraction, Opinion Words Extraction, KNN.