19 January 2026 in IEEE Access
The increase in digital communications has led to changes in the nature of documents that are often encountered in document examinations. Whereas document examiners use physical and chemical characteristics to authenticate documents and examine hypotheses about the age, origin, alterations, and signatures, digital documents cannot be subjected to such methods. This has led to calls for alternative approaches to digital document examination. In this paper, we present a comparative study that explores the determination of document origin, particularly, the identification of the creator tool type and the applicability of structural-based and machine-learning file type identification methods to this problem. To generate a baseline for the comparison of our methods, we implemented several machine learning algorithms that have most commonly and successfully been used for file type identification. These methods are then applied for tool type identification for digital documents to evaluate their relevance and effectiveness in this case. The paper focuses on the identification of creator tool types for Portable Document Format (PDF) documents due to their prevalence. We present our approach for identifying the tool used to create a document and evaluate our results against baseline results from file type identification. Our result shows that many of the structure-based analysis methods and machine learning algorithms used in file type identification can be extended to tool type identification in forensic digital document examination. The paper provides recommendations for adopting this extension and is the first work to explore the comparison of file type identification methods for creator tool type identification.