<![CDATA[In today’s world most of the people use social networking sites such as Twitter. They share their opinions and their views. through these media. Grouping these users will help us in different ways such as product recommendation, opinion mining, characterization of users based on their way of expressing their feelings. In this work, we present a technique to group the users based on the textual contents of the tweets. This technique is based on an unsupervised approach of machine learning that is clustering. A method is presented for representing the users using vector space model and TF-IDF weight scheme. K-means algorithm is employed for grouping the users using cosine distance as a distance measure. For the evaluation of this method, we construct a Twitter user dataset by using the Twitter application programming interface (API). A new technique is also proposed for characterization of the clusters formed. The experimental results are promising and from the study, it is found that the users in the clusters formed could be well defined by using the proposed cluster characterization technique.]]>
