Abstract: In today’s e-commerce world or in competitive market structure, lots of analyzed data is required for betterment of services, probability calculations, predictions, business decisions and summary of market reputation etc. This analysis is obtained through the detail summary of customer feedbacks and product reviews etc. To analyze this kind of data, opinion mining techniques are used. Hence for fine grained output from opinion mining, word alignment model and patterns of sentences are discussed in this project. Getting opinion words and opinion targets is the important and significant tasks. After detecting the opinion target and opinion word one important task is to determine relation between them. From the review it is clear that it is achieved using word alignment technique. Word alignment model is heavy tasks hence to balance the load and gear up the execution process partial supervised technique is used and syntactic patterns are used for it. Opinion relation graph is also formed to get proper analysis using Hill-Climbing algorithm. This opinion mining concept is also merged in web based application that recommends merchandise on the basis of analysis of product reviews. This web based system will be having feedback facility to analyse the web based facility. The product will be recommended as per the user requirement. Regular expression will reduce tasks of sentence comparison and analysis of other unnecessary data process.

Keywords: Opinion mining, opinion target, opinion word, WAM, co-ranking.