This study introduces a systematic web mining approach to collecting and analyzing data from extremist forums. Given the covert and dynamic nature of these online spaces, existing methodologies are inadequate. The proposed approach addresses practical challenges in identifying and collecting extremist forum content, leading to the creation of a database with 110 U.S. domestic extremist forums and over 640,000 documents. The study also explores multimedia usage, participant distribution, and posting activity within these forums, demonstrating the effectiveness of the approach and providing a valuable resource for understanding extremist movements.