Facebook to the Big Data Rescue!

Daniel Tunkelang had an interesting post on LinkedIn about Facebook search, welcoming Facebook to the ‘club’ of Human-Computer Information Retrieval. Facebook this week, as you’ve probably read, announced that they will launch a special search engine that takes into account all your connections, posts, comments, and those of others you can access, to serve you with answers to queries in ‘human language’. Facebook and search may be viewed with skepticism, as they don’t have a great image when it comes to privacy (also read this), but still, they may be on to something. From the demo, it seems several things are coming together here: one is to make sense of all the social data that you as a user already have access to. Not really Big Data, but interesting nonetheless, as it allows you to (hopefully quickly) zoom in on the answer to your query (‘single guys who are friends of a friend’ is mentioned as an example). The second is that after Apple Siri and Google Now, we now see Facebook take a stab at some sort of human-language interfacing. Perhaps we _are_ really getting closer to artificial intelligence after all! Or, as Tunkelang argued, closer to ‘human-computer information retrieval’: where people and computer work together to get to the best results. He in turn refers to Gary Marchionini, who proposed the following goals for these HCIR systems:

• Bring people closer to the information they need, especially to the meaning.
• Not only deliver the relevant documents, but also provide facilities for making meaning with those documents.
• Increase user responsibility as well as control.
• Be engaging and fun to use.

And these four bullets may very well describe what Big Data needs most: the human approach, the fun factor, the feel for what it is like to probe interesting datasets. The only way to create future data-scientists, or data-based-managers is to infect a whole generation with some sort of data savvy. As of yet, the role model of Google didn’t serve us well, as it often returns too much, out of context and (increasingly?) spam. Perhaps Facebook can help consumerize Big Data?