RENEWABLE BEHAVIORAL PATTERN GENERATION AND ANONYMOUS USER RECOGNITION METHOD AND SYSTEM

Abstract

This invention relates to a computer-implemented method and system that performs anonymous user recognition by utilizing behavioral data derived from user interactions during learning processes. Within the scope of this invention, user interaction data is acquired and preprocessed locally, behavioral characteristics are extracted from this data, and these extracted behavioral characteristics are converted into a comparable, fixed-size common representation across systems. This common representation is then subjected to a non-reproducible transformation process to convert it into a behavioral pattern that does not directly contain identifying information and is not suitable for reproducing raw behavioral data.The behavioral pattern is structured in two layers: a core pattern and a contextual pattern. Candidate recognition scores are calculated based on these patterns, and continuity verification is performed by evaluating the direction of change and consistency of behavioral patterns obtained from different time periods. The system also includes an adaptation structure that locally updates pattern weights, thresholds, and confidence measures by utilizing verified and false recognition results. Thus, users can be recognized through renewable and non-reproducible behavioral patterns without directly using personal information or sensitive biometric data.