How Recorded Future's prediction machine works

Recorded Future is explaining how its complex semantical search technology works and makes predictions and forecasts. Is Recorded Future the next Google, but more?
Written by John Dodge, Contributor

Recorded Future, an under the radar start-up company whose software can forecast trends and make predictions, is emerging from the shadows.

Since I first wrote about the new company on Feb. 22, I've gotten two e-mails from founder and CEO Christopher Ahlberg. At that time, the company disclosed little detail, but enough to make it an interesting SmartPlanet topic. The first e-mail links to a fascinating white paper that describes in considerable detail how Recorded Future works: Here's an excerpt:

Ahlberg from his LinkedIn photo

"We strive to provide tools which assist in identifying and understanding historical developments, and which can also help formulate hypotheses about and give clues to likely future events. We have decided on the term “temporal analytics” to describe the time oriented analysis tasks supported by our systems."

Recorded Future is trying to take Google a step further by analyzing semantics and sentiments to come up with predictions, trends and forecasts. Whether that could put a dent in demand for pricey reports done by research companies and mere humans remains to be seen.

How Recorded Future works.

In concept, Recorded Future is similar to so-called machine readable data products that Dow Jones, Thomson Reuters, Psydex and Selerity are creating in the financial industry.

They analyze the news to  predict how it will affect individual stocks or markets. I just wrote did an explainer on machine readable news (it could have a sexier name...) at Securities Industry News. And Recorded Future appears to be a close cousin to business intelligence (BI), but claims to be more forward-looking and capable of examining world events, not merely corporate and financial numbers, according to Ahlberg.

"The semantic text analyzes needed to extract entities, events, time, location, sentiment etc. can be seen as an example of a larger trend towards creating “the semantic web. We believe the future focus of business intelligence will be all about looking outside corporations and generating data and analytics for decision making based on the world, not just historical enterprise data. This is Recorded Future.”

It's an intriguing idea and is best visualized in the "Terrorism Analysis" video below which is dramatized with Middle Eastern music (the Middle East must be where terrorism comes from, right?). It shows a complex query using such search terms, among others, the timing between between bombings and comments made terrorism's top PR guy, al-Zawahiri. And it employs Google Earth to show travel patterns of key terrorism and counter-terrorism figures.

It's very slick, but in the end, the video fails to show the useful predictions that comes out on the results side. As close as it got to that was "We can also visualize networks of entities, relationships and related events over time...." Ok, how does that help me, exactly?

I'm guessing you keep plugging in search terms relevant to a particular topic until useful patterns, trends, predictions or visualizations emerge that helps your decision making.

The white paper describes in great detail how Recorded Future's technology works and comes up what it calls "Futures" which I assume are the payoffs.

"Futures are a way of storing analytic questions and having Recorded Future monitor them with respect to the continuous flow of data from the world."

The second Ahlberg e-mail that I just recieved invited me to subscribe via e-mail to these so-called Futures based on particular topics to me. Maybe, the "Futures" will show me what this does.

Follow me on Twitter.

This post was originally published on Smartplanet.com

Editorial standards