I have been asked to take a closer look at the Recorded Future site. A number of posts will follow that will track my interaction with the site and company.
The idea of a temporal analytics engine much intrigued me because time is usually the most difficult aspect of understanding data. That dimension is strongly complicating when we examine the past, because we often do not understand the contexts involved when the data was gathered. And in the future ... it is even more difficult to understand the contexts that will exist. Classic demand forecasting, practiced by every company juggles these issues to get the best results that it can. History is then be balanced against prediction to construct a useful model.
Recorded Future connects together a number of threads from predictors of the future, often stated in inexact human language, to predict aspects of the future. These are then visually presented to the analyst or executive to provide a solid basis for context rich understanding of a prediction. What will be particularly interesting is an exploration of how this kind of analysis can bridge the gap between classical, largely quantitative forecasting methods and more qualitative analysis of past and future events.
What I have seen so far is very impressive. Here is an example from an area I have worked in: Researching a Brand using Recorded Future. This short video example uses the IPad device and it's market as an example.
One thing to note that the demonstration shows. Prediction in time is dependent on relationships in time. And it is also strongly dependent upon networks of entities that influence changes. Recorded Future addresses and visualizes both
Would appreciate getting reports of experience with this idea in comments or privately. I am about to start a personal examination of it's use.
More posts on this concept will follow.