Freebase – Creating the Semantic Web

A couple of weeks ago I have got my invitation for Metaweb’s product Freebase and have played around with it a little bit since then. Freebase was launched in March and as you can see from the logo it’s still in alpha.

What is Freebase?

So what is Freebase about? Basically it is a collection of structured information on all kinds of topics and is connecting them with each other. At the moment a lot of information is based on Wikipedia articles while more and detailed information is contributed by users. If you promptly associate structured information with semantic web you are on the right track. Semantic web means that information is not only understood and interpreted by humans but more importantly by machines as well.

The image on the left describes how information is organised in Freebase; I have chosen the band Coldplay as an example.

First there are domains which are some rather broad categories that are grouping multiple types which can be seen as some sort of sub-categories. Topics are collections of different properties; certain values are assigned to them.

Now what’s really cool about Freebase is that once you start adding more content it automatically creates new topics and properties. For example I have added Subvert Central as a record label (type) as well as all the recording artists like Fracture & Neptune (topic); there was a new type (musical artist) created for Fracture & Neptune with Subvert Central being a label they have recorded for. So just one information creates many new ones.

Also multiple types can be assigned to just one person. Just have a look at the entry for David Lynch:

Conclusion

Freebase is a lot of fun to play with and combining Web 2.0 methods – users add content – to create a big database of semantic information is certainly a very interesting approach. However I don’t know if it will be as successful as Wikipedia because I doubt many people will really care that Freebase has a different approach (semantic web) to build a big database; they probably don’t want to add the same information again. It could also face vandalism and spam which might discourage people from contributing to it. Hopefully the Freebase creators are vigilant.

All data is licensed under a CreativeCommons license. There is also an API available which should make it interesting for developers, too.

If you want to test Freebase yourself leave a comment here; I have five invitations to give away.