Clickstream based user persona recognition
Investor: Telekom Slovenije
The team: principal investigator Andrej Košir, researchers Nejc Kozjek, Matevž Kunaver
The project aimed at the development, implementation and verification (feasibility study) of an unsupervised user persona identification algorithm based on remote controller click-stream.
We developed a novel user persona identification algorithm analyzing short sequences of user events. The core idea was a classifier where each processing step estimates its own reliability and in this way, the reliability of persona identification at each new click is estimated.