6 million spam tweets: A large ground truth for timely Twitter spam detection

Abstract

Twitter has changed the way of communication and getting news for people's daily life in recent years. Meanwhile, due to the popularity of Twitter, it also becomes a main target for spamming activities. In order to stop spammers, Twitter is using Google SafeBrowsing to detect and block spam links.Despite that blacklists can block malicious URLs embedded in tweets, their lagging time hinders the ability to protect users in real-time. Thus, researchers begin to apply different machine learning algorithms to detect Twitter spam. However, there is no comprehensive evaluation on each algorithms' performance for real-time Twitter spam detection due to the lack of large ground truth. To carry out a thorough evaluation, we collected a large dataset of over 600 million public tweets. We further labelled around 6.5 million spam tweets and extracted 12 light weight features, which can be used for online detection. In addition, we have conducted a number of experiments on six machine learning algorithms under various conditions to better understand their effectiveness and weakness for timely Twitter spam detection. We will make our labelled dataset for researchers who are interested in validating or extending our work.

A Survey of Spam Detection Methods on Twitter

“…Gao et al [82] propose a tweet-based spam detection approach based on the social degree of the tweet's sender, the history of interaction, the size of the cluster, the average time interval, the average number of URL in tweets, and the unique number of URL in tweets. Chen et al [83] present a real-time spam detection method for Twitter based on 12 lightweight features which are extracted from a dataset contains 6.5 million spam tweets. The features they consider detecting spam on Twitter are age of the account, the number of followers, the number of following, the number of likes the account received, the number of the account's lists, the number of tweets of the account, the number of retweets of the tweet, the number of hashtags used in the tweet, the number of mentioned users in the tweet, the number of URLs used in the tweet, the number of characters used in the tweet, and the number of digits used in the tweet.…”

Using supervised machine learning algorithms to detect suspicious URLs in online social networks

“…These can then be deployed directly into a machine learning algorithm. Table I shows examples of common features that have been used in previous studies [21] [11].…”

“…They do this by conducting spam campaigns that make their "fake" accounts connect with other fake accounts, increasing the follower and following numbers [22]. The majority of previous studies [23][24] [21] begin by collecting data using the Twitter streaming API 3 . Multiple features are then extracted and different feature sets utilized.…”

“…Similar to how security researchers study the attacks, spammers and hackers investigate detection systems; therefore, they can change user properties, content or the distribution mechanism to bypass certain restriction or detection rules [20]. For example, a study of detecting spam on Twitter [21] recommended that the number of followers is one of the highest discriminative power features. The feature's discriminative power has been increasingly weakened though by spammers making their accounts more popular.…”

Detecting Hacked Twitter Accounts based on Behavioural Change

“…Gawale and Patil implemented a system to detect malicious URLs on Twitter (Gawale and Patil, 2015). Chen et al (Chen et al, 2015) evaluated the ability of spam detection of various machine learning algorithms. They found that other classifiers tend to outperform Naive Bayes and SVM on spam detection.…”