Behavioral analysis and classification of spammers distributing pornographic content in social media

Abstract

Social spam is a huge and complicated problem plaguing social networking sites in several ways. This includes posts, reviews or blogs containing product promotions and contests, adult content and general spam. It has been found that social media websites such as Twitter is also acting as a distributor of pornographic content, although it is considered against their own stated policy. In this paper, we have reviewed the case of Twitter and found that spammers contributing to pornographic content follow legitimate Twitter users and send URLs that link users to pornographic sites. Behavioral analysis of such type of spammers has been conducted using graph-based as well as content-based information fetched using simple text operators to study their characteristics. In the present study, about 74,000 tweets containing pornographic adult content posted by around 18,000 users have been collected and analyzed. The analysis shows that the users posting pornographic content fulfill the characteristics of spammers as stated by the rules and guidelines of Twitter. It has been observed that the illegitimate use of social media for spreading social spam has been spreading at a fast pace, with the network companies turning a blind eye toward this growing problem. Clearly, there is an immense requirement to build an effective solution to remove objectionable and slanderous content as stated above from social networking websites to promote and protect public decency and the welfare of children and adults. It is also essential so as to enhance public experience of genuine users using social media and protect them from harm to their public identity on the World Wide Web. Further in this paper, classification of pornographic spammers and genuine users has also been performed using machine learning technique. Experimental results show that Random Forest classifier is able to predict pornographic spammers with a reasonably high accuracy of 91.96 %. To the best of our knowledge, this is the first attempt to analyze and categorize the behavior of pornographic users in Twitter as spammers. So far, the work has been done for identifying spammers but they are not specifically targeting pornographic spammers.

Human-Machine Collaboration for Content Regulation

“…Recognizing this, some scholars have begun developing automated (often, ML based) solutions to automate certain aspects of content regulation [28,42,45,68,94,108]. For example, researchers have proposed computational approaches to identify hate speech [18], pornography [98] and pro-eating disorder content [15]. Wulczyn et al created a ML classifier trained on human-annotated data to identify personal attacks in online discussions on Wikipedia [112].…”

Term Based Semantic Clusters for Very Short Text Classification

“…To our knowledge, invoice text classification has not been investigated previously1. Work related to text classification of short texts has been applied for microblogs (Singh et al, 2016;Ren et al, 2016;Missier et al, 2016), email subject classification (Alsmadi and Alhami, 2015) and spam detection (Bahgat et al, 2016).…”

Improved Micro-Blog Classification for Detecting Abusive Arabic Twitter Accounts

“…Spammers create and stockpile social media accounts, especially on Twitter because of its simplicity to create new accounts due to weak account opening and verification mechanisms. Spammers use these accounts to launch spamming campaigns that contain profanity, curse words, adult content, promotion of child pornography and exploitation, and harassment [13]. Spammers then disseminate targeted Twitter spam by exploiting weaknesses of the internet censorship and content filtering systems that use the blacklisted keywords, blacklisted URLs, and blacklisted spamming words [14].…”

“…In [13], Singh et al used six filthy keywords in Twitter searches to collect a pornographic spammer dataset. The dataset contains more than 73 thousand tweets and more than 18 thousand users.…”