Published Paper Details:

Online social networks (OSN) have greatly improved communication, information exchange, and enjoyment in modern society. However, because of their accessibility and anonymity, OSNs have provided a favorable environment for a variety of harmful practices like spamming, trolling, fake news, and astroturfing. One of the primary threats to OSNs is socialbots, which are computer programs that perform various illicit activities. To address the threat of socialbots, researchers have been developing various detection methods. One of the latest and most advanced methods is SBRidAPI, which stands for SocialBot RID (Rapid Identification) using deep API learning. The goal of SBRidAPI is to detect socialbots by analyzing a user's behavior on OSNs.SBRidAPI models a broad range of profile, temporal, activity, and content information for user behavior representation using deep learning techniques. Profile information includes user's name, age, location, and other similar data, whereas temporal information is about the frequency and timing of user activities. Activity information includes the type of activities, such as likes, comments, and shares. Content information includes the text, images, and videos shared by the user. SBRidAPI represents profile, temporal, and activity information as sequences in order to analyze the sequential nature of this information that is supplied to a two-layers stacked BiLSTM. Deep CNN is fed content data in order to analyze the text content and learn the visual characteristics of images and videos. Once SBRidAPI has analyzed a user's behavior, it assigns a score that reflects the likelihood of the user being a socialbot. The user is labeled as a socialbot if their score is higher than a threshold that is then used to compare scores. SBRidAPI is the first method that jointly models a complete collection of profile, temporal, activity, and content information for user behavior representation, making it an effective tool for identifying socialbots on OSNs.