Introduction to EmbedX
EmbedX is a large-scale embedding training and inference system developed using C++. It has been adopted by 12 different business sectors and utilized by over 30 teams within the organization, with more than a hundred successful deployments. The system has proven its effectiveness and performance in various applications, including recommendation systems, search engines, payment services, and risk control.
EmbedX has been implemented in numerous platforms and services such as WeChat Moments, WeChat Video Account, WeChat Search, WeChat Pay, WeChat Security, Tencent News, YingYongBao (App Store), QQ Music, JOOX Music, Tencent Classroom, Tencent's navigation platform, and Tencent's anti-fraud systems, achieving remarkable performance improvements and results.
For a more detailed overview, you can refer to the detailed introduction.
The academic paper on EmbedX was published in the PVLDB'2023 journal. If you want to cite the paper, you can use the following citation:
@article{10.14778/3611540.3611546,
author = {Zou, Yuanhang and Ding, Zhihao and Shi, Jieming and Guo, Shuting and Su, Chunchen and Zhang, Yafei},
title = {EmbedX: A Versatile, Efficient and Scalable Platform to Embed Both Graphs and High-Dimensional Sparse Data},
year = {2023},
volume = {16},
number = {12},
url = {https://doi.org/10.14778/3611540.3611546},
journal = {Proc. VLDB Endow.},
pages = {3543–3556}
}
Models and Evaluation
EmbedX boasts the implementation of several impressive models:
- Graph Models that handle billions of nodes and hundreds of billions of edges.
- Deep Ranking and Retrieval Models with tens of billions of samples and features.
- Joint modeling of graphs with deep ranking and retrieval models, handling billions of nodes, edges, and samples.
For detailed information, check the section on models and evaluations.
Getting Started Quickly
To begin using EmbedX, the following resources are available:
Using the System on a Private Cluster
- For standalone usage, refer to the single-machine guide.
- For distributed usage, consult the distributed usage guide.
Online Inference
Instructions for online inference can be found in the online inference guide.
Auxiliary Tools
EmbedX comes with various auxiliary tools including:
Contributing
Those interested in contributing to the project can find more information here.
Frequently Asked Questions
A comprehensive list of FAQs is available here.
For further questions or to contact the developers directly, refer to the Authors page.