Thu 13 Sep 2007
I’ve been poking around an interesting commercial initiative called “Freebase“, an open access / open licensed (using the Creative Commons attribution license) web-based data sharing system developed by Metaweb. Metaweb is a commercial enterprise, and according to their FAQ they plan on making money through some sort of fee structure on using their API (translation for archaeologists: an interface enabling machine-to-machine communication). Here’s a link to other blogger reactions and with lots of interesting discussion of Freebase.
I haven’t had any luck finding out how Freebase works, or what its underlying architecture is like. Given the shape of the Metaweb logo (triple lobes), I can only guess they have an RDF data-store (a big database of RDF-triplets). We’ll have an opportunity to learn more shortly, because Robert Cook of Metaweb has kindly agreed to speak about these efforts in our Information and Service Design Lecture series (at the UC Berkeley School of Information).
(Editing note: Here is a much more complete description of Freebase’s conceptual organization. )
However, my first impressions of surfing through Freebase remind me lots of some of the data structures we’ve been using in Open Context, which is based on the OCHRE project’s ArchaeoML global schema (database structure). For example, Freebase seems to emphasize items of observation that have descriptive properties and contextual relationships with other items. Open Context works just like that, but, being designed for the field sciences and material collections, Open Context assumes observations have some spatial relationships with one another (especially spatial containment). The overall point is that these systems offer data contributors tremendous flexibility in how they organize and describe their observations, while still enabling interoperability and a common set of tools for exploring and using multiple datasets. It’s a way of sharing data without forcing people into inappropriate, rigid or over specified standards.
Freebase looks more flexible in this regard (being designed for a wider set of applications). Freebase clearly has lots more professionalism in design and execution, and has an incredibly interesting API. It’s also great to see tools for data authors to share schemas (ways of organizing and describing datasets). All this shows you what great talent and venture capital funding delivers, and I’m duly impressed (and maybe a little jealous)!
We’re just now looking at RESTful web services for Open Context, and Freebase may offer an invaluable model / or set of design parameters for opening up systems for machine-to-machine interactions. In fact, making Open Context “play well” with a powerful commercial service such as Freebase would offer great new opportunities for our user community (choices of interfaces and tools).
Archaeology is a broad and diverse discipline, and making sure archaeologists can easily move data between different tools (blogs, online databases, and visualization environments like Google Earth) is an important need. We should take a serious look at systems like Freebase to make sure we’re best serving our community when we build such “cyberinfrastructure” systems.
BTW, anyone is welcome to help work with us on an archaeological web-services project. Open Context, unlike Freebase (which is a service built on a commercial product), is open sourced and you can get the source code here. It might be fun to come up with interesting ways to connect Freebase with Open Context.