Cross-Device Tracking: Matching Devices and Cookies

TL;DR

This paper presents a semi-supervised machine learning approach for cross-device user identification, addressing the challenge of fragmented user identities across multiple devices, and demonstrates competitive performance in a major challenge.

Contribution

It introduces a novel semi-supervised learning method for matching devices and cookies, advancing cross-device tracking techniques.

Findings

Achieved third place in ICDM 2015 Drawbridge challenge

Proved effectiveness of semi-supervised methods in cross-device identification

Improved accuracy over baseline approaches

Abstract

The number of computers, tablets and smartphones is increasing rapidly, which entails the ownership and use of multiple devices to perform online tasks. As people move across devices to complete these tasks, their identities becomes fragmented. Understanding the usage and transition between those devices is essential to develop efficient applications in a multi-device world. In this paper we present a solution to deal with the cross-device identification of users based on semi-supervised machine learning methods to identify which cookies belong to an individual using a device. The method proposed in this paper scored third in the ICDM 2015 Drawbridge Cross-Device Connections challenge proving its good performance.