Trajectory-User Linking Is Easier Than You Think

TL;DR

This paper shows that simple heuristics based on visit patterns are highly effective for Trajectory-User Linking, achieving up to 85% accuracy with minimal data and scaling to over 100,000 users, challenging prior assumptions about the task's difficulty.

Contribution

The study demonstrates that straightforward heuristics outperform complex models in TUL and introduces a scalable non-parametric classifier for large datasets.

Findings

Single check-in per trajectory predicts user identity up to 85%.

Non-parametric classifier scales TUL to over 100,000 users.

Visit patterns are highly unique, simplifying TUL.

Abstract

Trajectory-User Linking (TUL) is a relatively new mobility classification task in which anonymous trajectories are linked to the users who generated them. With applications ranging from personalized recommendations to criminal activity detection, TUL has received increasing attention over the past five years. While research has focused mainly on learning deep representations that capture complex spatio-temporal mobility patterns unique to individual users, we demonstrate that visit patterns are highly unique among users and thus simple heuristics applied directly to the raw data are sufficient to solve TUL. More specifically, we demonstrate that a single check-in per trajectory is enough to correctly predict the identity of the user up to 85% of the time. Moreover, by using a non-parametric classifier, we scale up TUL to over 100k users which is an increase over state-of-the-art by three orders of magnitude. Extensive empirical analysis on four real-world datasets (Brightkite, Foursquare, Gowalla and Weeplaces) compares our findings to state-of-the-art results, and more importantly validates our claim that TUL is easier than commonly believed.