This document summarizes a research paper that evaluates different classification methods for detecting spam users in social bookmarking systems. It tests naive Bayes and k-nearest neighbor classifiers on user data represented using three information retrieval models: Boolean, bag-of-words, and TFIDF. The best results were achieved using naive Bayes with a Boolean user representation, accurately classifying 97.5% of users. K-nearest neighbor worked best across all three representations, with over 96% accuracy using TFIDF. The study aims to automatically detect spam users through supervised machine learning techniques.