Unstructured human information makes up 90 percent of all known Big Data, and it's growing exponentially. For example, with more than one billion users, there are 700 Facebook updates per second. People tweet 340 million times per day, and there are 105 Google Plus updates per second. On a daily basis it is estimated that we collectively produce 2.5 quintillion bytes of data, and the growth rate is so high that 90 percent of all information ever created was produced in the past two years alone.

For a marketer, this unstructured human information is a goldmine. If a marketer can tap into the customer clicks and conversations happening every second and derive meaning from it, he or she can better understand each customer and deliver an optimized, personalized experience to that customer across their preferred channel and medium.

But actually doing this is the rub. The challenges are around volume and variety. Until recently, the quantity and speed of data has meant that by the time you are able to analyze it, it was no longer relevant. You can deal with problems of scale related to structured data by throwing increasing amounts of hardware and software at it. But human information in the form of call center conversations, social media interactions, video, images, web content, emails, and texts, doesn't fit cleanly into a database like structured data does. To understand and act on human information, you need software that can recognize concepts and ideas in this ocean of data.