🤖 AI Summary
This paper addresses the author verification (AV) task in digital text forensics, where existing methods suffer from high computational complexity, poor interpretability, and a lack of grounding in cognitive science. To overcome these limitations, we propose a lightweight syntactic modeling framework grounded in cognitive linguistics, treating individual syntactic patterns as behavioral biometrics. We introduce the λG metric—the likelihood ratio between an author-specific syntactic model and a population-level syntactic model—to enable intuitive and robust cross-document authorship attribution. Our approach integrates multi-granularity syntactic feature extraction, maximum likelihood estimation, and statistical likelihood ratio computation. Evaluated on 12 benchmark datasets, it significantly outperforms seven baselines—including multiple deep learning models—especially under low-resource, few-shot settings. The method delivers strong discriminative power, high interpretability, and practical applicability.
📝 Abstract
Authorship Verification (AV) is a key area of research in digital text forensics, which addresses the fundamental question of whether two texts were written by the same person. Numerous computational approaches have been proposed over the last two decades in an attempt to address this challenge. However, existing AV methods often suffer from high complexity, low explainability and especially from a lack of clear scientific justification. We propose a simpler method based on modeling the grammar of an author following Cognitive Linguistics principles. These models are used to calculate $lambda_G$ (LambdaG): the ratio of the likelihoods of a document given the candidate's grammar versus given a reference population's grammar. Our empirical evaluation, conducted on twelve datasets and compared against seven baseline methods, demonstrates that LambdaG achieves superior performance, including against several neural network-based AV methods. LambdaG is also robust to small variations in the composition of the reference population and provides interpretable visualizations, enhancing its explainability. We argue that its effectiveness is due to the method's compatibility with Cognitive Linguistics theories predicting that a person's grammar is a behavioral biometric.