This doctoral thesis investigates methods for detecting plagiarism in academic documents beyond textual content. It presents approaches for citation-based, image-based, and math-based plagiarism detection, and develops a hybrid plagiarism detection system combining different detection methods. The thesis evaluates the proposed methods on relevant datasets and analyses their effectiveness at identifying various forms of plagiarism. It finds that the citation-based, image-based and math-based detection methods can successfully retrieve source materials for confirmed cases of plagiarism involving these non-textual elements. The hybrid system integrates the different detection approaches to provide a unified interface for comprehensive plagiarism analysis.