Social networking is a very popular communication tool where opinions are shared by millions of users thereby generating tremendous data every day. Adjectives in this data represent emotions such as good, amazing etc. for positive nature and poor, bad, inferior for negative nature. Also emoticons such as ( for happy feeling and ( for sad feeling etc are used which are necessary to be treated along with adjectives to know blogger’s expression. Opinion mining is generally performed by matching words to a dictionary or a corpus that contains static set of words. However if the user uses new words which are not present in the existing dictionary, those words may be neglected. To overcome this shortcoming, this paper proposes a novel algorithm concerned with auto addition of new words at runtime thus making the dictionary dynamic. The algorithm not only gives ratings for adjectives but also for emoticons. The polarity of these newly added words will be defined by considering the polarity of its neighboring words. The analysis shows that novel algorithm gives more accurate ratings than the previous findings. These accurate ratings can help users to choose the products wisely before buying thereby benefiting the vendor as well as users.

Opinion mining sentiment analysis emoticons analysis social network mining dynamic opinion mining