Current Visitors:
VecCity: A Taxonomy-guided Library for Map Entity Representation Learning
1. Introduction
VecCity is an open-sourced and standardized benchmark compatible with various datasets and baseline models.
2. The VecCity Library
The overall framework of VecCity. For a given dataset, we first construct atomic files from multi-sourced city data by extracting and mapping entities (e.g., POIs, road segments, land parcels, etc.) and auxiliary data into corresponding atomic files. MapRL models encodes various entities in a unified configuration, which facilitates joint processing for various downstream tasks.
Overview of the VecCity library
The main features of LibCity can be summarized in three aspects:
- Unified: LibCity builds a systematic pipeline to implement, use and evaluate traffic prediction models in a unified platform. We design basic spatial-temporal data storage, unified model instantiation interfaces, and standardized evaluation procedure.
- Comprehensive: 42 models covering four traffic prediction tasks have been reproduced to form a comprehensive model warehouse. Meanwhile, LibCity collects 29 commonly used datasets of different sources and implements a series of commonly used evaluation metrics and strategies for performance evaluation.
- Extensible: LibCity enables a modular design of different components, allowing users to flexibly insert customized components into the library. Therefore, new researchers can easily develop new models with the support of LibCity.
The reproduced traffic prediction models are categorized in the below table.
The implemented models in LibCity
3. Contact Us
VecCity is mainly developed and maintained by Beihang Interest Group on SmartCity (BIGSCITY). Welcome to visit and use our repository for more details. Your suggestions and contributions are very important to us! If you have any questions about the library, please raise an issue in our repository.
4. Cite
If you find VecCity useful for your research or development, please cite the following paper.
Wentao Zhang, Jingyuan Wang, Yifan Yang, and Leong Hou U. 2025. VecCity: A Taxonomy-Guided Library for Map Entity Representation Learning. Proc. VLDB Endow. 18, 8 (April 2025), 2575–2588. https://doi.org/10.14778/3742728.3742749