Implementation of Association Rule Mining Algorithms with Big Data for Identification of User Pattern in Social Networks

Abstract

Information Technology and Communication (ITC) technologies are recently growing much newlinefaster. Due to this a number of services and applications have become online and offer benefits newlineto consumers. These data-driven applications are a large and growing source of information, newlinewhich provides us with various kinds of valuable information and patterns. Among them, social newlinemedia and e-commerce platforms are popular applications, where a significant amount of data newlinewill be contributed by the users. However, there is a large amount of data published but without newlineany content monitoring system. The content monitoring system can control the contents and the newlinecontent promoters. But, social media analytics is a big data analytics problem, which cannot deal newlinewith the normal data mining techniques. Therefore, the proposed work is focused on the study of newlineBig Data technology, which provides data analytics services for large amount of social media newlinedata. newlineThe proposed work utilizes data analytics technology for contributing into two critical social newlinemedia issues. First, the problem is associated with fake account detection. Fake accounts are one newlineof the most crucial problems in social media, because, using the fake accounts the malicious newlineusers are performing some unsocial tasks such as phishing, distribution of hate speech, and newlineothers. In this context, we proposed a fake account classification model based on association rule newlinemining and user profile attributes. That system works on the social media fake account dataset newlineobtained from GitHub, which consists of 2818 instances of profiles and consists of 34 attributes. newlineThese attributes are distributed among two kinds of samples i.e. fake and legitimate. Here we newlineneed to process a total of 95,812 items for extracting the association rules. These rules are further newlineoptimized with a developed filtering algorithm and can be used for the classification of fake newlineaccounts based on the attributes of the Twitter profile newline