Subtle Patterns in Complex Events

I have now seen a number of live demonstrations of Recorded Future and am increasingly intrigued by the subtle complexity that is revealed when we can do something that is more that just simple search.

I have been involved in a number of projects that dealt with focused and unfocused competitive intelligence. These are most useful when they utilize human domain expertise to do searches and look ups - and them combine what is found into a useful outline that synthesises the results.

What is often lost are precisely those subtle patterns that search cannot find. Human experts may find it, but they are often not given the right data. Or if they have the right data, cannot quantitatively compare results. Combinatorically too, there may be too many patterns to compare. Time domain aspects of a problem are hard to manage. It is also problem being addressed by interactive, visual business intelligence.

In the Recorded Future Blog, an instructive post on how they are addressing this problem, read the entire post:

Searching Complex Events to Reveal Subtle Patterns: Currencies, Terrorism, and Corporate Actions
A very neat feature in Recorded Future is the ability to do advanced temporal queries looking for complex events involving the relationship of two patterns. We might want to systematically explore or monitor for patterns such as:

» Two people speaking about the same subject within a short period of time
» A person speaking leading to certain actions
» A company action preceeding another action – potentially “shady” combinations
Since the data in Recorded Future comes from thousands and thousands of sources, such trends become particularly interesting to explore .... "