The rapid growth of the Internet has made extracting information from Web pages critical. Pages contain main content as well as noisy blocks like navigation, copyright, ads, and hyperlinks, which hinder clustering, classification, and retrieval. Challenges increase with regional languages. This paper presents a novel pixel-map-based approach for content extraction and knowledge creation for illiterate users, combining pixel attributes, pattern matching, statistical models, and Artificial Neural Networks. The goal is concept-based term analysis at sentence and document levels rather than single-term analysis. © 2015 Elsevier B.V., All rights reserved.