In a previous post, we considered how facebook Graph Search could potentially be used within the automotive context, covering possible use cases etc – Graph Search in general clearly offers huge potential and this is also true of the automotive context.
In Carmesh, we were interested in exploring what is possible with facebook data and, in particular, whether some of the Graph Search capabilities are available to external applications.
The first point to note is that facebook Graph Search is not available via facebook’s API. Graph Search is – for now – something that is only available within the facebook ecosystem and even then, it is only being rolled out on a gradual basis, so only a fraction of users have access to it right now. Obviously, facebook have invested heavily in Graph Search and want to leverage it themselves in the first instance; only then might they consider making it available outside of facebook. Graph Search per se is not available to third party developers.
However, facebook does offer some useful and related functionality to third party applications. Although Graph Search itself is not available, the facebook Graph API has a very useful search capability which can be leveraged. This search function has a couple of different capabilities which are worth noting:
- get ‘likes’ of given user : this can be used to obtain a set of items on the facebook Graph that the person likes, some of which can be location based;
- get locations relevant to the given user: this can be used to obtain a set of locations within a specified area which are of interest to the user – these can be places where the user has checked in, where the user or their friends have been tagged, etc
The order in which the above are returned is of some import: the set of likes are returned in chronological order, ie the order in which the user liked them. This does make it a bit more difficult to work with them, eg it’s not possible to query likes by location or type (cafe, restaurant, for example). Also, the set of locations is decided by facebook and they do not specify how this is determined, although it is personalized and does attempt to prioritize places of interest to the user. As with the set of likes, they do not make it possible to perform a location search using type.
The above two sets of data can be used meaningfully. However, it is clear that facebook’s priority right now is not to make this data set very flexible for third party developers – it is their data after all! Also, facebook does seem to be getting a little more protective of its value and is putting more emphasis on ensuring that third party developers give something back to facebook in return for the use of the data.
The challenge, then, is to make use of the data in a meaningful way without imposing excessive load on facebook, while also giving back to the facebook ecosystem. Oh, and integration with Google’s data, of course, but perhaps that’s a subject for another post…