Abstract: The review texts with subjective sentiments on the Web are valuable for many applications, since they expressed the opinions, attitudes and standpoints of users. The opinion mining technictue process these subjective texts specially to generate useful opinion summarization and knowlcdgcs automatically. The background and application of opinion mining was introduced firstly and then the sate of the art of opinion mining was presented in five subtasks: sentiment polarity identification of words, coarse sentiment classification, fine grade opinion mining and summarization, opinion retrieval and language resources and application systems; The difficulty and trend of opinion mining was concluded finally.

Key words: Sentiment polarity, Sentiment classification, Opinion mining, Opinion retrieval