Place Deduplication with Embeddings

TL;DR

This paper addresses the challenge of deduplicating places in large, multi-source place graphs by developing a novel embedding-based pipeline that significantly improves over existing methods.

Contribution

It formulates the place deduplication problem, explores related tasks, and proposes a systematic, data-driven embedding pipeline with novel techniques that outperforms current state-of-the-art solutions.

Findings

The proposed method achieves higher accuracy than existing approaches.

The embedding pipeline effectively captures place similarities.

Significant improvements demonstrated on Facebook's place graph data.

Abstract

Thanks to the advancing mobile location services, people nowadays can post about places to share visiting experience on-the-go. A large place graph not only helps users explore interesting destinations, but also provides opportunities for understanding and modeling the real world. To improve coverage and flexibility of the place graph, many platforms import places data from multiple sources, which unfortunately leads to the emergence of numerous duplicated places that severely hinder subsequent location-related services. In this work, we take the anonymous place graph from Facebook as an example to systematically study the problem of place deduplication: We carefully formulate the problem, study its connections to various related tasks that lead to several promising basic models, and arrive at a systematic two-step data-driven pipeline based on place embedding with multiple novel techniques that works significantly better than the state-of-the-art.