International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 06 Issue: 12 | Dec 2019 www.irjet.net p-ISSN: 2395-0072 © 2019, IRJET | Impact Factor value: 7.34 | ISO 9001:2008 Certified Journal | Page 739 An Effective Analysis of Anti Troll System using Artificial Intelligence Aishwarya Gaikwad 1 , Mrunmayee Patil 2 , Sarang Patil 3 , Mayura Rane 4 , Dr. Shwetambari Chiwhane 5 1,2,3,4 B.E. Student, Dept. of Computer Engineering, NBN Sinhgad School of Engineering, Ambegoan, Pune- 411041, Maharashtra, India 5 Professor, Dept. of Computer. Engineering, NBN Sinhgad School of Engineering, Ambegaon, Pune – 411041, Maharashtra, India ---------------------------------------------------------------------***---------------------------------------------------------------------- Abstract - Online harassment has been on the rise ever since rampant boom in social media. Trolling is just another form of bullying that found its roots over the web. Certain anti-troll measures should be taken to deal with these issues and avoid promoting it further. Nowadays it has become a trend on social media to spew toxic hate. Some manual measures such as ignoring or blocking the trolls have been in use, but with the rise in the number of trolls, it needs more of an automated approach. Few social media platforms block trolls based on their set of troll words, however trolls resist these anti-troll systems by intentionally misspelling or other cunning methods. This paper discusses implementation of anti-troll using machine learning and artificial intelligence to provide a smarter troll detection system that adapts to current and updated trolling sense. Key Words: Sentiment Analysis, Artificial Intelligence, Anti Troll System, Twitter Sentiment Analysis 1. INTRODUCTION Trolling on social websites has become very common activity nowadays. It is a huge issue in virtual world. As bullies have no restrictions from anyone they can easily get away after trolling a person. There is a need to create an application software to detect such kind of hazardous trolling and warn that bully so that he would think twice before doing such actions. Many companies have taken steps regarding trolling activities. There are very few software available which could detect foul words and simply block them but in troll detection system, our respective software system needs to have a clear understanding of sentences and clear language used by the troll. In this paper we are exploring various “Anti trolling systems” and different functionalities used in that system. This paper also suggests some machine learning algorithms and sentimental analysis used in anti-trolling systems. We are using Twitter social networking platform as an API for detecting our trolls using various methodologies. This paper also discusses about current functionalities and architecture used for anti-trolling systems and acquire a vision of trolling free internet. 2. RELATED WORK 2.1 Literature Survey on Sentimental analysis The main objective of carrying out sentimental analysis is to detect trolls. During this process few important steps such as: 1. Task definition 2. Annotation guidelines 3. Data collection and annotation. Under task definition few terms are used: Repetitiveness- trolls send a large number of troll messages. Destructiveness- troll messages express negative sentiments to sow discord Deceptiveness: troll messages may be deceptive to achieve their objective of creating discord.