M Vinay Sai et al., International Journal of Emerging Trends in Engineering Research, 8(7), July 2020, 3345 - 3350 3345 ABSTRACT With the huge increase in the number, speed, and variety of customer information (such as customer-generated information) in online interpersonal organizations, people have worked hard to construct better methods to collect and inspect such large information. For example, social robots have been used to perform scientific management of robots and provide customers with improved management attributes. Even so, harmful social robots are used to spread false data (for example, false news), which may bring real results. Therefore, identifying and evacuating harmful social robots in online informal communities is crucial. The latest discovery technology of resentful social robots undermines the quantitative focus of its behavior. These highlights are easily imitated by social robots. This leads to a reduction in the accuracy of the investigation. This article describes an epic strategy for identifying malicious social robots, which includes two key options, depending on the possibility of snapshot stream grouping and semi-hosted binding changes. The technology not only investigates the possibility of changes in the click stream of customer behavior, but also takes into account the temporal highlights of the behavior. Key words: Online social network, social robots, Customer behavior, semi-supervised clustering. 1. INTRODUCTION Data Mining is a method that organizations use to turn raw data into usable information. Using algorithms to scan trends in vast volumes of data, corporations can learn more about their clients and create more successful content campaigns, boost revenue and lower expenses. Data mining relies on effective data collection, storage, and computer processing. Supermarkets are well-known customers of information mining methods. Many grocery stores provide customers with free membership cards, so that they can get preferential prices that non-individuals cannot enjoy. Such cards make it easier for shops to keep track of who ordered what at what price and when. Once the information has been disaggregated, the shop would be able to utilize this information and provide consumers with coupons customized and their purchase preferences, and chose whether to reduce the products mentioned or market them to the maximum degree practicable. When an organization uses only selected data (it cannot explain a general set of examples) to illustrate a particular theory, information mining may be the cause of concern. Data mining is a method of analysis designed to analyze data (typically massive volumes of data, generally relevant to industry or market, also known as 'big data') to locate common patterns and/or machine relationships between variables, and then to validate the findings by extending the trend found to a new subset of data. A definite aim of information mining is prescient, and prescient information mining is the most commonly accepted method of information mining and the most immediate application for industry. The knowledge mining technique includes three phases: (1) the primary inquiry (2) the model layout or sample validation with clarification/check, and (3) the arrangement. 2. LITERATURE SURVEY Some of the issues commonly associated with anonymity online are that it impedes a sense of social obligation, as has been demonstrated by a great deal of fake news online. Despite the lack of prompts for identification in cyberspace, people still leave fragments of textual identities behind. In this report, we recommend using methods for the stroke examination to better classify individuals based on writing style. They incorporate a rich range of features of the type, including terminology, grammar and form, content-specific attributes and functionality [1], [5]. We have also developed Write prints technology for anonymous identification and similarity detection. Write prints is a technology based on the Karhunen-Loeve transform, which uses sliding windows and mode interrupt algorithms in combination with a separate author-level feature set. On a test bench containing four online datasets spanning different fields, the Write prints technology and extended feature set were evaluated: email, instant messaging, feedback comments, and program code. 3. EXISTING METHODOLOGIES We are improving the current architecture for a Scalable and Robost Truth Discovery (SRTD) scheme to tackle the above three challenges. In fact, using a rational approach, the SRTD scheme mutually quantifies both the authenticity of data and the credibility of statements. Using Function Queue in a Detecting and Analyzing the Malicious Social Bots by using Data Mining and Naïve Bayesian Classifier M Vinay Sai 1 , Gandharba Swain 2 , K Hari Kishore 3 1 M.Tech Student, Department of CSE, Koneru Lakshmaiah Education Foundation, Vaddeswaram, Guntur, A.P, India, vinaysaimuppalla@gmail.com 2 Professor, Department of CSE, Koneru Lakshmaiah Education Foundation, Vaddeswaram, Guntur, A.P, India 3 Professor, Department of ECE, Koneru Lakshmaiah Education Foundation, Vaddeswaram, Guntur, A.P, India ISSN 2347 - 3983 Volume 8. No. 7, July 2020 International Journal of Emerging Trends in Engineering Research Available Online at http://www.warse.org/IJETER/static/pdf/file/ijeter75872020.pdf https://doi.org/10.30534/ijeter/2020/75872020