COMPARATIVE ANALYSIS OF SPAM DETECTION TECHNIQUES IN SOCIAL NETWORKS

Abstract

The proliferation of social media platforms has brought about a surge in spam
content, posing significant challenges to user experience, trust, and security. In response,
various machine learning (ML) algorithms have been employed to detect and mitigate spam
activities. This study presents a comparative analysis of spam detection techniques across
five prominent social media platforms: Facebook, Twitter, Instagram, LinkedIn, and
Messenger. Five distinct ML algorithms, namely Logistic Regression, Support Vector
Machine (SVM), K-Nearest Neighbors (KNN), Decision Tree, and Random Forest, were
implemented to evaluate their efficacy in detecting spam content on these platforms. The
performance of each algorithm was assessed based on accuracy metrics, with the results
visualized through bar plots for comprehensive comparison. The project involved collecting
and preprocessing data from each social media platform, ensuring representation of diverse
spam characteristics prevalent across different networks. Features such as content type,
frequency, user engagement, and network-specific attributes were considered during data
preprocessing to enhance model effectiveness. Subsequently, the data were divided into
training and testing sets for model training and evaluation. Logistic Regression, known for its
simplicity and interpretability, demonstrated competitive performance across all platforms,
effectively discerning spam from legitimate content. SVM, leveraging its ability to handle
high-dimensional data, exhibited robust performance particularly on platforms like LinkedIn
and Twitter. KNN, relying on similarity metrics, showcased notable accuracy in identifying
spam on Facebook and Messenger. Decision Tree, with its intuitive decision-making process,
yielded promising results on Instagram. Random Forest, an ensemble method, consistently
delivered strong performance across all platforms, leveraging the diversity of decision trees
to combat spam effectively. The bar plots illustrating the compared accuracies of each
algorithm provided valuable insights into their relative strengths and weaknesses across
different social media platforms.