Litcius/Paper detail

Spam Detection On Social Media Platforms

R. Abinaya, Bertilla Niveda E., P Naveen

20202020 7th International Conference on Smart Structures and Systems (ICSSS)25 citationsDOI

Abstract

With the increased quality of online social platforms, spammers have come up with various techniques to lure users into accessing malicious links. This is done by generating spam on the comment section of various social media networks. In this paper, we've taken YouTube comments as the dataset and performed spam YouTube comment detection. The current methods to stop spammers include using tools such as Google Safe Browsing which help to detect and also block irrelevant spam on YouTube. Although these tools help in blocking harmful links, it fails to secure the users in real-time scenarios. Thus, many different approaches have been applied to form an environment that is spam free. A few of them are solely supported user-based options whereas others are based on YouTube content. We have assessed our answer with four completely different algorithms based on machine learning, namely - Logistic Regression, Decision Trees Classifier, Random Forest, Ada Boost Classifier and Support Vector Machine. With Logistic Regression, an accuracy of 95.40% is possible, surpassing the current solution by roughly 18%.

Topics & Concepts

Computer scienceSocial mediaSupport vector machineRandom forestClassifier (UML)Decision treeLogistic regressionMachine learningSpambotWorld Wide WebBlock (permutation group theory)Blocking (statistics)SpammingArtificial intelligenceThe InternetComputer networkGeometryMathematicsSpam and Phishing DetectionSentiment Analysis and Opinion MiningHate Speech and Cyberbullying Detection