Organizing What the World Knows about the Future Staffan Truvé, PhD CTO
Jan 21, 2015
Organizing What the World Knows about the FutureStaffan Truvé, PhD
CTO
2
Why Recorded Future?
• The web is loaded with predictive signals
• Temporal analysis needed to reveal them –
search is not enough
• Allows for shifted focus of (business) intelligence:
Outwards and Forward, not Inwards and Backward
Web Loaded with Predictive Signals
04/10/2023 3
Silicon Valley executives head to Vail, Colo. next week for the annual Pacific Crest Technology Leadership Forum
The carrier may select partners to set up a new carrier as early as next month
“2010 is the year when Iran will kick out Islam. Ya Ahura we will.”
“... Dr Sarkar says the new facility will be operational by March 2014...”
Drought and malnutrition hinder next year’s development plans in Yemen...
“...opposition organizers plan to meet on Thursday to protest...”
“Excited to see Mubarak speak this weekend...”
“According to TechCrunch China’s new 4G network will be deployed by mid-2010”
“Strange new Russian worm set to unleash botnet on 4/1/2012...”
Some questions you cannot aska search engine about
• Which heads of state visited Libya in 2010?
• What pharma companies are releasing new
products in the first quarter of 2012?
• What do French bloggers say about the
earthquake in Haiti?
From Unstructured Text to Analyzable Data
Recorded Future Data
STRUCTURE
ME
TR
ICS
TIM
E
STRUCTURE
A scene has to have a rhythm of its own, a structure of its own.
Michelangelo Antonioni
10
STRUCTURE
• ENTITIES:
persons, places, products, technologies, companies, …
• EVENTS:
meetings, travels, acquisitions, earnings calls, product
releases, natural disasters, elections, protests …
• ONTOLOGIES:
what entities exist, and how can they be grouped, related
within a hierarchy, and subdivided according to
similarities and differences:
Geography, World Leaders, Corporations and officers,
Technology Areas, …
12
Port au Prince
TIME
If you can look into the seeds of time and say,
which grain will grow, and which will not,
speak then to me. (Macbeth, Act 1 Scene 3)
14
TIME
• Publishing Time vs. Event Time
• Event time: a period in time when an event has
occurred or is expected to occur
• Derived from publishing time +
natural language processing of text
METRICS
Don't pay any attention to what they write about
you. Just measure it in inches.
Andy Warhol
METRICS
• Numeric attributes derived from text fragments,
documents, and larger context
• Momentum – “media buzz”
• Calculated based on aggregate information about an entity
/ event
• Based on relative frequency of occurrence, credibility of
sources, co-location with other entities etc.
• Sentiment (positive, negative, deceit, uncertainty, …)
Momentum for Michael Jackson 2008-2011
Negative sentiment for Muammar al-Gaddafi, 2011
FACTS
20
21
FACTS
• Fact = “the mentioning of an event in a text”• Event type Acquisition
• Attributes (entities) Acquirer: Google, Acquired:Motorola Mobility
• Event time 2011-08-15 – 2011-08-15
• Source New York Times
• Momentum 0.8
• Sentiment Positive: 0.3, Negative: 0.1
• Text fragment Google Inc. and Motorola Mobility
Holdings, Inc. today announced that
they have entered into a definitive
agreement under which Google will
acquire Motorola Mobility for …
• More than 3 billion facts in Recorded Future index
SEARCH
22
STRUCTURE
ME
TR
ICS
TIM
E
Querying the index
STRUCTURE
ME
TR
ICS
TIM
ELAST 24 MONTHSPRODUCT RELEASE -- APPLE
RESULT SET ORDERED BY METRIC
PREDICTIONS"Thus, what enables the wise sovereign and the good general to
strike and conquer, and achieve things beyond the reach of ordinary
men, is foreknowledge."
(from The Art of War by Sun Tzu, Section 13)
SENSORS FOR CURRENT STATE + MATHEMATICAL MODEL PREDICTIONS
Predicting through algorithmic crowdsourcing
As of Sept. 15 2011
Wikipedia:The southwestern summer monsoons occur from June through September.
Learning how the world works
• Predicting trading volume
• Predicting stock returns from sentiment
• Predict volatility by future events
• Single blog impact analysis
http://www.predictivesignals.com/
Using Recorded Future Data to Trade
31
Unlock the predictive power of the web!