DecKG: Decentralized Collaborative Learning with Knowledge Graph Enhancement for POI Recommendation

TL;DR

A novel decentralized collaborative learning with knowledge graph enhancement framework for POI recommendation (DecKG), where instead of directly uploading interacted items, users generate desensitized check-in data by uploading general categories of interacted items and sampling similar items from same category.

Abstract

Decentralized collaborative learning for Point-of-Interest (POI) recommendation has gained research interest due to its advantages in privacy preservation and efficiency, as it keeps data locally and leverages collaborative learning among clients to train models in a decentralized manner. However, since local data is often limited and insufficient for training accurate models, a common solution is integrating external knowledge as auxiliary information to enhance model performance. Nevertheless, this solution poses challenges for decentralized collaborative learning. Due to private nature of local data, identifying relevant auxiliary information specific to each user is non-trivial. Furthermore, resource-constrained local devices struggle to accommodate all auxiliary information, which places heavy burden on local storage. To fill the gap, we propose a novel decentralized collaborative learning with knowledge graph enhancement framework for POI recommendation (DecKG). Instead of directly uploading interacted items, users generate desensitized check-in data by uploading general categories of interacted items and sampling similar items from same category. The server then pretrains KG without sensitive user-item interactions and deploys relevant partitioned sub-KGs to individual users. Entities are further refined on the device, allowing client to client communication to exchange knowledge learned from local data and sub-KGs. Evaluations across two real-world datasets demonstrate DecKG's effectiveness recommendation performance.

Figures (4)

• Figure 1: The overview of DecKG. a) Step 1: Knowledge graph is pretrained on the server side without the sensitive user-item interaction data. b) Step 2: At the initial stage, client upload the desensitized check-in data to the server, while server assign neighbors and partition the KG into $KG_i$ for user $u_i$. c) Step 3: Users are trained with the private check-in data combining the sub-KGs, and the knowledge from neighbors in client-client communication.
• Figure 2: Performance of different hyperparameters on Beijing.
• Figure 3: performance of DecKG integrating to other DecPOIs on Beijing.
• Figure 4: Performance of different DecKG variants on Beijing.