Clustering Model for Microblogging Sites using Dimension Reduction Techniques

Related Works

Clustering of microblogs is a very significant challenge in the social media analysis domain, which has resulted in some quality research works over the years. Miyamoto, Suzuki & Takumi (2012) used a fuzzy neighborhood model for analyzing the tweets, and then used two kernel-based methods along with pair wise constraints for clustering the tweets. Streamcube is a hierarchical spatiotemporal hashtag clustering system which has been proposed in Feng (2015). In their research, a spatiotemporal hierarchy inspired by quad-tree and data cube is used to perform hashtag and clustering according to a divide and conquer strategy at the most last level of the steps. Then an event ranking algorithm is used that helps users to identify local and burst events (which is one of the main problems that clustering solves). Perez-Tellez et al. proposed a method of clustering the microblog texts associated to any organization for online reputation management (Perez-Tellez, Pinto, Cardiff, & Rosso, 2011). They introduced the unsupervised TEM-Full and TEM-FULL+F methods to enrich term representation of tweets that is later used for clustering which has enabled clustering algorithms such as K-means, to obtain superior performance in clustering.