The NYT wrote a good overview.
And Tim ORielly has two thoughtful posts on the new site (one, two). O'Rielly describes Freebase as "the bridge between the bottom up vision of Web 2.0 collective intelligence and the more structured world of the semantic web."
Here's what it is all about. Freebase is cross between free-form user contributed content (think Wikipedia) and a structured database (think schemas and SQL).
A screen shot makes this more concrete:
Here, the entry for "O'Rielly" has a series of fields allowing users to add data to in a structured format. A "company" (eg O'Reilly) would have fields for "stock ticker", "founding date", "location", "revenues", etc. (I suspect, but am not sure, that users are the ones associating fields with entries or types -- vs. ontological experts, as in the approach taken by Cyc).
Like Wikipedia, users will provide much of the data. Freebase also slurped large databases -- including Wikipedia -- to start the the ball rolling. Unlike Wikipedia, the user-contributed/user-reviewed/user-edited data is structured, so machines can do interesting things with it.
Even Freebase itself benefits from this semantic structure: when the field "Location Founded" gets populated for a company (for example, "Newton Massachusetts" is the "Location Founded" for "O'Rielly"), the entry for the place "Newton, Masschusetts" immediately becomes aware that O'Reilly was founded there. In Wikipedia, in contrast, cross-references are built and reviewed by humans.
Take a look at the tagset for Google Base, and you can see Google is heading down the same path towards semantic meaning.
Call it "Web 3.0". Call it "the semantic web." Whatever the label, it will revolutionize information online.
Today, machines understand which web pages connect to which other pages via hyperlinks. However, the modest link doesn't provide much meaning as to why the pages relate -- is the link an endorsement? an attack? a citation? a sarcastic reference? -- and web pages are too large of chunks to serve as atomic elements of meaning.
When we provide the machines with the structure to understand relationship between data elements, all sorts of interesting things will become possible very quickly.
Taking the narrow perspective, all these developments will strongly impact online retail, as they will reshape how consumers find and choose products and services. Taking the broad perspective, the semantic web will change the fabric of online data, and how human knowledge is stored and used.
Keep a close eye on Freebase and GoogleBase -- the semantic web promises to be a wild ride over the next few years.