Face-ing your future

A lot of you are absolutely aware of what Web 2.0 means and what change that has brought about. Among the plethora of changes that it continues to offer, the rise of Social Networks and the way it permeates into each of our lives is no longer something new. To be fairly honest, articles talking about the rise of Social Networking and the way people interact on Facebook these days sounds trite and contrived. Yes they exist, but how do they transform computing in general? I haven't found a lot of articles that talk about what is in store for the next two years in terms of computing and in terms of what to do with the wealth of information that these networks dish out to us and what techniques are used to tackle them.


While a single glance at a Facebook page, one can really predict the type of the user. *This applies to any human who just looks at the profile and the Wall updates, along with the frequency of updates*. While I am not going to discuss on the updates that focus on how a single great looking female's inanity has been appreciated or how a certain celebrity's foolishness has been applauded, I am certainly looking at how these obscure status messages can turn into potential business goldmines.

Coming to the crux of the situation. A seemingly innocuous status message like "My dream!" along with a picture of a "Porsche Cayenne" gives an indication that the poster is concerned about or even likes that car. If Chaos Theory can argue that the fluttering of a butterfly's wings can cause an earthquake on the other side of a world, surely a status message can prompt a lot of businesses to latch on to the opportunity. While not everyone might want a "Cayenne" but surely a lot of us might want Xboxes, Wiis, Cameras, Mobile Phones, Kindles and what not? Enter Natural Language Processing. For decades AI researchers have talked about how NLP can be used to understand text. While most of these projects were really academic and served as interest to a lot of scientists and researchers, their true test comes out in this age of the unstructured data. A system with even 50% of accuracy, when run on a set of 100,000 status messages can give you a substantial idea on what people are talking about. Why then, these businesses can utilize this to directly talk to the user, probably come up with offers and go one step ahead in the marketing of the product.

Even as we are talking a lot of people are looking at this problem and are figuring out ways to solve them. One of the greatest challenges in building such a system has been the way in which such a system can be show-cased to the client. While Human Beings can be inane, nothing can take away their brains from them. So where are the stumbling blocks to processing this?

1)How do you understand the domain that the user is speaking( Even if the domain is resolved, how can the machine evolve and learn? Can parsers co-exist with ontologies? Is an ontology based parser probable? If we claim that the brain has an internal wiring and then assuming that ontology is the poor cousin of the brain, can we parse?

2) How do you present it as a sell-able product? What kind of examples do you show case the client? This is definitely the biggest challenge. The challenge arises from the fact that the client here is definitely presented with a virtual goldmine hidden in some treasure chest and the problem is that he does not know how do get to the goldmine. More often than not the client might get the wrong assumption while looking at an application. The way he tests it should be on the same lines that the program was tested. How can both the business user and the application developer align their thoughts? Very critical as it makes or breaks your product.

3) Finally how do you scale the system?. Scalability is always the question in many applications, but in my case I'd rather say, how easy is it that the system learns new patterns. Can it apply the patterns from one domain and use the same to extend it to another domain? Would that be a worthwhile route to take?

Loads of questions, very little and diverse answers, the future, probably is not bleak, but then there are tough times ahead for products which don't embrace the power of Social Networks.