Since the rapid growth of social media content has promulgated in digital space, the amount of data that has been generated via social media got multiplied every day. Moreover, the social media contents are very complicated and loosely coupled for integration. Besides it paves the tough path for capturing the data and analyzing the data for future decision making process. In order to facilitate this difficult operation, we have proposed here the clustering technique to effectively cluster the data in lesser amount of time. This paper implements the PSO algorithm with the modification required to match the Twitter data streams. The outcomes clearly depict the performance of the PSO algorithms and shows that there is an increase in number of particles in the larger selection of data streams. The proposed work has taken two different approaches to cluster the Twitter data streams using the modified PSO algorithm. First, it has used to select the centroids based on the number of clusters given by the user. Second, refine the cluster for giving the high convergence to the selected data sets.