Recently Facebook launched beta version of Graph Search service for English-language users from the U.S. Graph search is a search engine that provides personalized results, taking into account the an individual’s relationships, ultimately returning objects within the Facebook Open Graph. Here, we consider how this may be used in the vehicular context.
Graph search supports queries for people, posts, events, places. For example, search queries such as “find friends who posted a picture from Las Vegas” or “find single males friends of friends who live nearby and come from China” are possible.
How does this differ from a standard Google search? It differs in two key aspects: firstly, as the nodes on the graph are typed entities, the searches can be more accurate in many cases. Secondly, and more interestingly, these results are personalized based on the user’s profile and connections, rather than search history.
This type of search is very sensitive to the individual performing the search: the results will differ quite significantly for different people. This is because everyone’s set of relationships over which the search is done are different.
As well as standard objects available in Facebook, the Open Graph API enables new objects related to the user’s activity to be defined and managed from different applications. Other apps may publish stories on behalf of the users, e.g. “John Doe had a test drive in the new Ferrari LeFerrari” or “John Doe had a spin in Lisa Smith’s new Mini Cooper S”. This gives external apps a huge added value – it can use the mix of facebook data and their own to provide additional service.
So, how can vehicular services make use of these new capabilities? The possibilities are endless! For example, we may recommend a restaurant or a social event for the driver based on his/her profile and current destination; moreover, we can additionally enhance the relevance of the results taking into account the correlation between user’s preferences and those of his/her friends. For example, when searching for a restaurant, the results could be weighted by friends that have similar food preferences, rather than weighting all friends equally. While searching for location based recommendations, we may also want to search not only within our friends, but people who actually live there.
As well as destination based searches, route based searches could be performed: this would enable us to find interesting places en route. In this way interesting resting, eating or filling stations could be proposed to the driver.
Using Facebook as the only source of data is limiting. Other data sources could also be combined to give better results. Tripadvisor is an obvious data source which provides excellent quality rating data and much greater detail on specific places en route. This could be easily combined with Graph Search to provide more information to the user.
In conclusion, taking into account the current nascent trend in the auto industry to open up their APIs to third party developers, integrating high quality comprehensive data on places, relationships, activities and personal interests, we can be very excited about the future of “on-board” applications, which will improve our experience both as a driver and as a passenger.