What the Hell is a Graph Search? The Graph Search Strikes Back

I know I’ve written about Facebook’s new Graph Search once before, but given what we’ve been studying recently, it seemed appropriate to revisit the topic. Plus, I recently gained access to the Graph Search Beta! (It’s precisely as creepy as everyone makes it out to be)

Facebook’s new Graph Search is meant to be a form of personalized search; it utilizes information from the user’s social network to tailor results to that user. Two different users entering the same query could get radically different results.

What struck me about Graph Search the second time around is its likely similarity, on a fundamental level, to traditional search engines. While Facebook keeps the algorithms that compose Graph Search heavily guarded, it is not hard to imagine that Graph Search could easily use an algorithm not entirely unlike the hub-authority computation, with the content being searched for as the “authority” and the user’s friends as the “hubs.” Like the hub-authority computation, potential authorities are determined by keyword search while the structure of links between hubs and authorities determine how highly authorities rank in the search results. The concept of influence in online social networks is not a new one, and it is analogous to the values assigned to various hubs in hub-authority computations.

So if the fundamental algorithm of the search isn’t different, than what is the point of Graph Search? Graph Search is importantly different from traditional search in two ways: 1) The set of data to be searched has been, heretofore, private. Vacation photos and check-ins are qualitatively quite different than what one might find in a Google image search or a Yelp review. This is content generated by the everyday life experience of users, rather than by content-providers jockeying for search result placement. 2) Hubs are not determined by the structure of the network, they are instead user-selected representatives of the user’s offline experience. The user’s offline life determines, to a large extent, the search results he or she gets.

This unique situation, that of an otherwise private dataset ranked by both keyword relevance and its relations to the user’s offline social relationships, makes me wonder how businesses will respond. Social media is the lifeblood of many businesses, and I’m curious what Graph SEO will look like.