Potential Applications of Machine Learning in Forensic Questioned Document Examination
Surbhi Mathur, Sumit Kumar Choudhary, Parvesh Sharma, Kashvi Sood, Vinay Aseri · 2025
The chapter discusses the potential applications of artificial intelligence and machine learning (AI–ML) techniques in forensic document analysis. The field of forensic document examination has been dominated by manual methods of examination worldwide in the absence of credible automation methods, which can offer evidence beyond a reasonable doubt, and thus the inherent challenges and limitations of manual examination very much exist. The AI–ML techniques that have revolutionized almost every aspect of human intervention, promises to be an effective tool for forensic document examination as well. The present chapter highlights the importance of automated document verification systems in detecting fake documents and reducing identity fraud. The development of machine-learning tools that can assess language structure and formation from threatening emails, messages and posts can aid in identifying suspects who try to hide their identities online. The chapter also mentions the use of AI–ML techniques in offline signature verification, multilingual handwritten numeral recognition and historical document analysis. It emphasizes the need for more tools with similar functionality to transform current procedures and techniques for document analysis. The chapter explores different aspects of forensic document (signatures) analysis with the help of the 4×4 grid method and runs a gradient structural and concavity features (GSC) system for verifying the nature of grid features with appropriate graphical information, including writer identification, handwriting authentication, text-line extraction and statistical writer ship analysis.