This document presents a critical review report on a course about algorithm design and analysis. It summarizes a paper that studied graph-based forecasting techniques for social networks. The paper generated a social graph from user activity data and used rule-based mining to predict future social behaviors. It found that this method was more efficient than the Apriori algorithm as it did not require multiple scans of the dataset. The real-world simulation achieved accuracy rates of over 80% for predictions made over 10 days. However, the review critiques that the paper's title was too broad and its impact was limited by only being cited 3 times in the last 7 years. Overall, the paper presented a good recommender system for social networks.