Enterprise text analytics is an exciting and powerful area gaining traction recently. When one moves on from purely departmental solutions to a perspective of an enterprise text analytics fabric, new possibilities emerge to empower a wide array of roles and functions.
Unstructured data is all around us: in news stories, web pages, journal articles, social media posts, patents, research reports, presentations, and a variety of other sources. These items are unstructured in that they don’t start out with a predefined, explicit schema or structure. Historically, these documents have been read by humans looking to find information relevant to their particular tasks or roles. In today's deluge, however, the need for scalable reading, repeatability, traceability, and speed has driven the advent of text analytics platforms.
When I went to Text Analytics World in San Francisco earlier this month, I was struck at how many of the presenters, particularly consultants, ended their talks describing future directions of text analytics as something that sounded so familiar. They described what would be possible once there's advanced maturity in ontologies, the breaking down of siloes, entity and relationship resolution by multiple methods, and automated linking of it all together into semantic network models of knowledge: flexible exploration of the relevant. They made it sound like a bit of a stretch, almost pie in the sky, but what they briefly described as this destination was curiously similar to what was shown concretely in the last presentation of the conference, my own.