Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Correlation-Attention Masked Temporal Transformer for User Identity Linkage Using Heterogeneous Mobility Data

Ziang Yan, Xingyu Zhao, Hanqing Ma, Wei Chen, Jianpeng Qi, Yanwei Yu, Junyu Dong

TL;DR

This work tackles user identity linkage (UIL) across heterogeneous mobility data by modeling spatio-temporal co-occurrences between cross-platform check-ins. It introduces MT-Link, a Correlation Attention Masked Transformer that uses a Spatial-Temporal Embedding Layer, a Temporal Transformer Encoder, a Correlation Attention Block, and an attention-guided Masked Transformer Encoder to retain co-occurrence points while suppressing noise. The approach yields significant improvements over state-of-the-art baselines on four real-world cross-platform datasets, with notable gains in Macro-F1 and AUC and robust ablation and sensitivity results. The method offers a scalable, noise-robust solution for UIL in heterogenous location-based networks, enabling more accurate cross-platform user matching and downstream analytics.

Abstract

With the rise of social media and Location-Based Social Networks (LBSN), check-in data across platforms has become crucial for User Identity Linkage (UIL). These data not only reveal users' spatio-temporal information but also provide insights into their behavior patterns and interests. However, cross-platform identity linkage faces challenges like poor data quality, high sparsity, and noise interference, which hinder existing methods from extracting cross-platform user information. To address these issues, we propose a Correlation-Attention Masked Transformer for User Identity Linkage Network (MT-Link), a transformer-based framework to enhance model performance by learning spatio-temporal co-occurrence patterns of cross-platform users. Our model effectively captures spatio-temporal co-occurrence in cross-platform user check-in sequences. It employs a correlation attention mechanism to detect the spatio-temporal co-occurrence between user check-in sequences. Guided by attention weight maps, the model focuses on co-occurrence points while filtering out noise, ultimately improving classification performance. Experimental results show that our model significantly outperforms state-of-the-art baselines by 12.92%~17.76% and 5.80%~8.38% improvements in terms of Macro-F1 and Area Under Curve (AUC).

Correlation-Attention Masked Temporal Transformer for User Identity Linkage Using Heterogeneous Mobility Data

TL;DR

This work tackles user identity linkage (UIL) across heterogeneous mobility data by modeling spatio-temporal co-occurrences between cross-platform check-ins. It introduces MT-Link, a Correlation Attention Masked Transformer that uses a Spatial-Temporal Embedding Layer, a Temporal Transformer Encoder, a Correlation Attention Block, and an attention-guided Masked Transformer Encoder to retain co-occurrence points while suppressing noise. The approach yields significant improvements over state-of-the-art baselines on four real-world cross-platform datasets, with notable gains in Macro-F1 and AUC and robust ablation and sensitivity results. The method offers a scalable, noise-robust solution for UIL in heterogenous location-based networks, enabling more accurate cross-platform user matching and downstream analytics.

Abstract

With the rise of social media and Location-Based Social Networks (LBSN), check-in data across platforms has become crucial for User Identity Linkage (UIL). These data not only reveal users' spatio-temporal information but also provide insights into their behavior patterns and interests. However, cross-platform identity linkage faces challenges like poor data quality, high sparsity, and noise interference, which hinder existing methods from extracting cross-platform user information. To address these issues, we propose a Correlation-Attention Masked Transformer for User Identity Linkage Network (MT-Link), a transformer-based framework to enhance model performance by learning spatio-temporal co-occurrence patterns of cross-platform users. Our model effectively captures spatio-temporal co-occurrence in cross-platform user check-in sequences. It employs a correlation attention mechanism to detect the spatio-temporal co-occurrence between user check-in sequences. Guided by attention weight maps, the model focuses on co-occurrence points while filtering out noise, ultimately improving classification performance. Experimental results show that our model significantly outperforms state-of-the-art baselines by 12.92%~17.76% and 5.80%~8.38% improvements in terms of Macro-F1 and Area Under Curve (AUC).

Paper Structure

This paper contains 19 sections, 8 equations, 5 figures, 3 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Problem Definition
  4. Methodology
  5. Spatial-Temporal Embedding Layer
  6. Temporal Transformer Encoder
  7. Correlation Attention Block
  8. Masked Transformer Encoder
  9. User Identity Linkage Layer
  10. Experiments
  11. Datasets
  12. Baselines
  13. Evaluation Metrics and Experiment Settings
  14. Experimental Results
  15. Ablation Study.
  16. ...and 4 more sections

Figures (5)

  • Figure 1: Check-ins of the same user on different platforms.
  • Figure 2: The overview of the proposed framework.
  • Figure 3: The impact of mask ratio and correlation attention block layers on MT-Link. Tw-Fb represents the Twitter-Facebook dataset, and Tw-Fs represents the Twitter-Foursquare dataset.
  • Figure 4: Co-occurrence visualization.
  • Figure 5: The time costs of MT-Link and other baselines.

Theorems & Definitions (2)

  • Definition 1: Check-in Point
  • Definition 2: Check-in Sequence