International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 08 Issue: 05 | May 2021 www.irjet.net p-ISSN: 2395-0072 © 2021, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 868 SMART FAKE NEWS PREDICTION USING MACHINE LEARNING FOR SOCIAL MEDIA Puja Sunil Erande 1 , Monika Dhananjay Rokade 2 1 PG Student, Department of Computer Engg., SPCOE, Maharashtra, India 2 Assistant Prof., Department of Computer Engg., SPCOE, Maharashtra, India ---------------------------------------------------------------------***--------------------------------------------------------------------- Abstract - Fake news are described with an intention to misdirect or to delude the reader. We have presented a response for the task for fake news, individuals are clashing if not by large poor locators of fake news. For this reason new system is generated for fake news identification. The result of this project determines the actual fake news detection for social networks using machine learning. Number of peoples having social media accounts such as facebook, whatsapp, twitter,etc. This social network is main source of news. Because of the wide effects of the huge fake news, individuals are clashing if not by large poor locators of fake news. While these systems are utilized to make an increasingly dynamic complete start to finish arrangement, we need to talk with to progressively troublesome cases where progressively solid sources and creators release counterfeit news. As, the goal of this model was to make an apparatus for recognizing the language plans that depict fake and certified news utilizing AI, AI and regular language preparing strategies. The results of this system demonstrate the limit with regards to machine learning and AI to be significant. We have developed a new system that gets many no of natural signs of genuine and fake news & also an application that guides in the representation of the classification choice. Key Words: Content modeling, Fake news detector, Fake news categorization, Stance detection, Machine learning, Social media, online fake news, twitter. 1. INTRODUCTION There are a number of people having profiles on social media platforms (SMPs) are growing, thus hiding their identity for malicious purposes. Over the last few years, online social networks have seen both the number of users and the amount of information shared explosively rise. Users may use these sources of messages to connect, share, discover and disseminate information. Some of those services provide social connections (Facebook and Twitter, for example). Others (YouTube and Flicker, for starters) are used for sharing content. One of the main research problems is determining what users do on such sites. System Uses Twitter's Social Network as our case study. To identify the document, numerous techniques were suggested, including rule-based, neural network, decision trees, and machine learning. There are also several machine learning-based tricks and classifications. The basic idea behind these strategies is to identify news types using a qualified classifier that can predict some of the predefined classes associated with a news category automatically. Nave Bayes employs the idea of chance. The parameter in Nave Bayes was taught by training the module with the Bayesian rule of probability. The performance of a system that represents a text document as a bag of words with each word considered independent of the others is primarily degraded. 2. HISTORY AND BACKGROUND According to [1] the event-based approach based on consumer curiosity used by LeMeNo for News Recommendation. The network of recommendations is focused on both current events and customer expectations. News articles are recommended using machine learning techniques such as grouping related articles, predicting their content, subject similarity, and keyword extraction. The system learns user preferences based on the amount of time spent reading a post, as well as the user-specified rates of interest in different subjects. In this day and age, where there are so many news sources to choose from, it's critical to develop a solution that can guide customers to relevant articles based on their preferences. To increase the likelihood of users recommending a related post, our architecture integrates several approaches to news recommendations. According to [2] Evaluates some of the most Machine learning techniques are commonly used to automatically identify Nepali data, particularly Naive Bayes, SVM and Neural Networks. The method is being experimented with a self-created Nepali News Corpus with 20 different categories and a total of 4964 posts, gathered online by crawling various national news portals. Functionality dependent on TF-IDF is derived to train and examine the models from the preprocessed documents. The classification pip. According to [3] Social Poisson factorization (SPF), a Probabilistic model incorporating social network information into a standard factorization method; SPF applies to the algorithmic suggestion a social aspect. It provides a robust method to test SPF data and shows that it outperforms rival methods on six datasets in the real world; data sources include a social reader and Etsy. According to [4] Privacy risks Similar to numerous emerging and influential automation patterns, including