Mining Based Design and Analysis of Social Spam Detection in Micro-blogging

Abstract

The web-based social networking becomes a valuable part of over life. Young clients can pay a significant amount of time on this social platform. The primary reason behind the time expense on the social media is to check the updates on the different area of interest i.e. politics, movies, and others. The updates on these domains are obtained on the basis of the trending topics. But sometimes the similar or duplicate topics are flooded on social media due to this un-necessary traffic, redundancy, and storage overheads increases. Keeping in mind the end goal need to identify the duplicate post on the social network applications and remove them is a better solution. By this inspiration a new data model using the big data mining is introduced in this work. The proposed data model contributes by accepting the online and offline data both. After that the three phase of pre-processing is performed on the data first the removal of stop words, removal of punctuations, and completion of abbreviations. The pre-processed data is further ranked on the basis of Jaccard similarity index. This ranked data is further used with the fuzzy c-means algorithm. The fuzzy c-means algorithm computes the different groups of the similar tweets. Thus in further for finding the similar tweets the synonyms based re-tweets are generated with the mutation methodology. Finally the hashes of all the data are computed and the similar hash value based tweets are removed. The implementation of the proposed method is finished on the idea of JAVA era and hadoop storage. Additionally after implementation of the proposed technique, the technique is compared with the similar technique on the basis of their precision and recall values. The computed results demonstrate the high degree of accurate duplicate data identification and their removal for the micro-blog data analysis.