Research at Freebase
Freebase is a practical, scalable, graph database of structured general human knowledge, created by Metaweb Technologies. It is designed to scale to a large number and diversity of users and data. Freebase allows public read and write access through an HTTP-based graph-query API for research, personal, and commercial purposes. Access is free and all data in Freebase has a very open (e.g. Creative Commons, GFDL) license.
We encourage academic research use of the the data in Freebase, for purposes such as data mining, machine learning, Web analysis, NLP, network/graph analysis, or any other interesting questions or problems you may want to investigate. Metaweb provides several services and information to support research, including:
Freebase API
Freebase can be directly accessed via its HTTP API using the Metaweb Query Language (MQL). This API supports all of the read/write capabilities of Freebase. Feel free to try it out!. Client code written in python can be found here.
A Sandbox For Experimentation
A snapshot of www.freebase.com is made each week (usually Monday afternoon) for experimental purposes and made available at sandbox.freebase.com. Sandbox is useful for testing/debugging Freebase applications, schema testing, and data representation exploration. Large reads/writes to the Freebase HTTP API should be tested here before trying them at www.freebase.com. Please feel free to contact us if you have questions about sandbox usage.
Bulk Datasets
We will be providing wholesale extracts of interesting large data sets. The first two of these are wholesale snapshots of all of the facts in Freebase and a processed Wikipedia extraction called WEX. The wiki markup has been transformed into clean XML and some structural information (e.g. sections, templates) has been extracted.
Bulk Processing With Hadoop
We are offering our Jython-based framework for writing map-reduce jobs on Hadoop as an open source project. Happy tightly embeds Jython into the Hadoop APIs, files off a lot of the sharp edges, and makes writing map-reduce programs a breeze. This is the 0.1 release, but we've been using Happy at Freebase for a while, so it is stable and full-featured.
Community And Support
There is an active community of Freebase users, and we encourage you to join us. Feel free to share with the world what you are doing with Freebase and ask questions when you need help. You can be part of the Freebase community in many different ways:
- Join the mailing lists for Freebase developers and data modelers,
- Check out the Freebase Blog to find hints, tricks, and technicial explanations.
- Browse the substantial Technical documentation.
- Try out some of the several interesting applications have been created using Freebase, many with source code available.
Contact
Freebase is created by the world for the world, and the links above should help you get started. However, if you have problems, questions, or ideas related to your research project that are not covered here or are very subtantial in scope, feel free to send mail to research@metaweb.com, and we will be happy to chat with you.
Publications
- Kurt Bollacker, Patrick Tufts, Tomi Pierce, Robert Cook, A Platform for Scalable, Collaborative, Structured Information Integration, AAAI-07, Sixth International Workshop on Information Integration on the Web

