This study introduces REST, a systematic methodology designed to extract useful information from security forums, specifically focusing on identifying threads related to security threats, malicious services, hacking information, and useful security experiences. REST addresses challenges such as defining relevant topics and dealing with the informal nature of forum text. It utilizes a multi-step weighted embedding approach to project words, threads, and classes into embedding spaces, assessing relevance and similarity. Evaluated on data from three security forums with 164k posts and 21k threads, REST demonstrates robustness to keyword selection and superior accuracy in categorizing threads (63.3-76.9%) compared to five other methods. This approach offers a user-friendly way to leverage forum information with minimal keyword input.