Attention Span For Personalisation

TL;DR

This paper introduces a framework for collecting user engagement data from event streams, extracting features like attention span, and demonstrates that recommendations based on attention span significantly improve click-through rates.

Contribution

It presents a novel framework for capturing and utilizing user engagement features, especially attention span, to enhance recommender system performance.

Findings

Attention span-based recommendations increased click-through rate by 340%.

The framework effectively extracts engagement features from event streams.

The approach improves personalization in recommender systems.

Abstract

A click on an item is arguably the most widely used feature in recommender systems. However, a click is one out of 174 events a browser can trigger. This paper presents a framework to effectively collect and store data from event streams. A set of mining methods is provided to extract user engagement features such as: attention span, scrolling depth and visible impressions. In this work, we present an experiment where recommendations based on attention span drove 340% higher click-through-rate than clicks.