Abstract: Mining opinion targets and opinion words which are extracted from online reviews are two most essential components for efficient opinion mining which is achieved by identifying opinion relations within words. Analysis done in this piece of work is an approach based on partially supervised alignment model. Then a co-ranking algorithm is used to develop a graph model to identify approximate level of confidence of each candidate and at the end of all opinion targets and opinion words are selected based on confidence level of the candidate.

Keywords: Sentiment Analysis, Opinion target, Opinion word, Word Alignment Model.