Methode, System und Computerprogramm zur Analyse von Benutzerverhalten

Details

Veröffentlichungs-Nr.

EP4730166

Anmeldetag

16. Oktober 2024

Veröffentlichung

22. April 2026

Rechtsraum

EP

IPC

G06F21/31, G06N20/00

Abstract

A user (41) behaviour analysis system in the form of user (41) interactions with an end device (12), comprising an input module (43) that receives signals from the end device's (12) sensors (42) during user (41) interactions with end device (12), a memory module (14) that records sensor (42) signals during user (41) interactions with end device (12), a comparison module (44) that uses a pre-trained machine learning model (30), and an output module (46) that provides the response from the comparison module (44). The system is characterised in that the comparison module (44) uses a pre-trained machine learning model (30) that responds to a set of signals from the sensors (42) of the device (12) during the interaction of the user (41) with the end device, representing at least two interactions (I, G) of the user (41) with the end device (12), to generate a response in the form of a similarity metrics (P) between interactions (I, G); furthermore, the comparison module (44) contains a pre-trained machine learning model (30) which is a machine learning model trained on interactions originating from a closed group of end device users (12) other than the current end user (41).

Anmelder

Firma

Naukowa i Akademicka Siec Komputerowa Panstwowy Instytut Badawczy

Land

🇵🇱 Polen