There is a discussion going on which you can follow on Techmeme about the post by Aditya Agarwal who is a Facebook employee ( more specifically a Facebook Tech Lead ) on whether Facebook really needs to write their own search engine.
Aditya Agarwal shares about how results are sorted using the concept of ’social proximity’ where “People closer to you in the graph—your friends and people in your networks—are likely to be more relevant to you and thus are ranked higher”.
While Facebook’s search engine will probably contain more juice behind the results that are returned, I am surprised that Aditya Agarwal uses only social proximity to highlight what he considers as Facebook’s search key differentiator which is the use of an individual’s place in the social graph to return results. The potential of the social graph in people search is so much more.
Currently, if I search with Basic Search using the phrase ’sex in the city’ the profiles returned would be those which contain the term ’sex in the city’ where the top results would be my friends, followed by everyone else in my network.
How does Facebook choose the order to display the rest of the profiles of those in my networks? I don’t know, but it would seem that whatever they use, it doesn’t use the social graph or use it well. Somewhere on the 6th page of the results, I saw a classmate from my high school. I haven’t added him yet as a friend on facebook, but one of my friends who I have added did. Which means, someone who was a friend in real life, and a friend of a friend on facebook’s social graph was being ordered less than some other people in the Singapore network (the only network I am on) who I don’t even know and can’t fathom the connection.
Using the social graph could produce more relevant results beyond social proximity. For example, if two individuals in my network have the phrase ‘rock climbing’ in their profiles, but one of them, say individual A has 300 friends in his other networks who have ‘rock climbing’ on their profiles while individual B has maybe only 10 such friends, then individual A’s profile should be higher up in the results. Does this seem familiar? Google uses a similar concept in establishing authority and relevancy as part of their PageRank technology.
Facebook has the potential to apply the social graph to produce much better results not just based on my position in the social graph (i.e. social proximity) but also based on other people’s position in the social graph and the links that create those relationships.
Facebook might soon be a marketing person’s dream tool if they are able to search and extract data about relationships from the social graph effectively. Imagine being able to identify who are the Connectors and Maven. Imagine being able to contact them to help spread a message and see how that message spreads across the newsfeed of the members on Facebook. Imagine being able to experiment in how to make a message sticky enough so that it can spread and reach a tipping point.
But before we get to all that, maybe their basic search needs some working on. I know a friend who likes a particular Singaporean movie “Money No Enough“. I tested the Advanced Search by keying in the phrase “no enough” in the Movies field. No results. I then tried the full movie title “Money No Enough” before getting my friend as a result. If Facebook really wants to be a Google-killer in people searching, they probably need to start allowing people to search without using exact phrases.
Aditya Agarwal ends his post by sharing how Facebook has big plans to improve their search in the upcoming months. I cannot wait to see how they leverage the social graph to do that.
Post a Comment