Application of Kullback-Leibler divergence for short-term user interest detection

TL;DR

This paper introduces a mathematical framework using Kullback-Leibler divergence to detect short-term user interests in recommender systems, addressing the limitations of traditional static preference models.

Contribution

It proposes a novel information-theoretic approach for real-time user interest detection based on item properties, improving reactive recommendation capabilities.

Findings

Framework effectively captures short-term user interest shifts

Enhances recommender system responsiveness to recent user activity

Based on fundamental information theory principles

Abstract

Classical approaches in recommender systems such as collaborative filtering are concentrated mainly on static user preference extraction. This approach works well as an example for music recommendations when a user behavior tends to be stable over long period of time, however the most common situation in e-commerce is different which requires reactive algorithms based on a short-term user activity analysis. This paper introduces a small mathematical framework for short-term user interest detection formulated in terms of item properties and its application for recommender systems enhancing. The framework is based on the fundamental concept of information theory --- Kullback-Leibler divergence.