This paper addresses the problem of identifying rumors and disinformers in microblogs, focusing on Twitter. The authors propose a framework that uses statistical models and a linear function of log-likelihood ratios to retrieve rumors from a large dataset of manually annotated tweets. They explore three categories of features: content-based, network-based, and microblog-specific memes. The experiments show that the proposed method achieves a Mean Average Precision (MAP) of over 0.95 in rumor retrieval and effectively identifies users who endorse rumors. The dataset, consisting of over 10,000 tweets, is the first large-scale public dataset for rumor detection and can be used to study misinformation and information diffusion in online social media. The authors also discuss future work on identifying new emergent rumors directly from Twitter data.This paper addresses the problem of identifying rumors and disinformers in microblogs, focusing on Twitter. The authors propose a framework that uses statistical models and a linear function of log-likelihood ratios to retrieve rumors from a large dataset of manually annotated tweets. They explore three categories of features: content-based, network-based, and microblog-specific memes. The experiments show that the proposed method achieves a Mean Average Precision (MAP) of over 0.95 in rumor retrieval and effectively identifies users who endorse rumors. The dataset, consisting of over 10,000 tweets, is the first large-scale public dataset for rumor detection and can be used to study misinformation and information diffusion in online social media. The authors also discuss future work on identifying new emergent rumors directly from Twitter data.