An effective assessment of cluster tendency through sampling based multi-viewpoints visual method

Social networks are the rich sources to people for sharing the knowledge on health-related issues. Nowadays, Twitter is one of the great significant social platforms to the people for a discussion on topics. Analyzing the clusters for the tweets concerning terms is a complex process due to the sparsity problem. Topic models are useful or avoiding this problem with derivations of topic clusters. Finding pre-cluster tendency is the major problem in many clustering methods. Existing methods, such as visual access tendency (VAT), cosine-based VAT (cVAT), multi viewpoints-based cosine similarity VAT (MVS-VAT) majorly used to access the prior information about clusters tendency problem. Solution of cluster tendency indicates the tractable number of clusters. The MVS-VAT enables the cluster tendency for the tweet documents effectively than other visual methods. However, it takes a higher number of viewpoints, thus requiring more computational time for the clustering of tweets data. Therefore, sampling-based visual methods are proposed to overcome the computational problem. Several standard health keywords are used for the extraction of health tweets to illustrate the effectiveness of proposed work in the experimental study.