Learning Causality for News Events Prediction Kira Radinsky, Sagie Davidovich, Shaul Markovitch Technion - Israel Institute of Technology.

Download Report

Transcript Learning Causality for News Events Prediction Kira Radinsky, Sagie Davidovich, Shaul Markovitch Technion - Israel Institute of Technology.

Learning Causality
for News Events Prediction
Kira Radinsky, Sagie Davidovich, Shaul Markovitch
Technion - Israel Institute of Technology
What is Prediction?
“…a rigorous, often quantitative,
statement, forecasting what will happen
under specific conditions.“ [Wikipedia]
“A description of what one thinks will
take place in the future, based on
previous knowledge.” [Online Dictionary]
Why is News Event Prediction
Important?
Event
Predicted event (Pundit)
Al-Qaida demands hostage A country will refuse the
exchange
demand
Strategic Intelligence
Volcano erupts in
Democratic Republic of
Congo
Thousands of people flee
from Congo
Strategic planning
7.0 magnitude earthquake
strikes Haitian coast
Tsunami-warning is issued
China overtakes Germany Wheat price will fall
as world's biggest exporter
Strategic planning
Financial investments
Outline
• Motivation
• Problem definition
• Solution
• Representation
• Algorithm
• Evaluation
Problem Definition: Events Prediction
Ev is a set of events
T is discrete representation of time
Prediction Function
, s.t.:
occurred at time
occurred at time
Outline
• Motivation
• Problem definition
• Solution
• Representation
• Algorithm
• Evaluation
Causality Mining Process: Overview
News Articles acquisition
• Crawling [NYT 1851-2009]
• Modeling & Normalization
Causality Pattern
Classification
• <Pattern, Constraint,
Confidence>
Event Extraction
• Tagging
• Dependency parsing
(Stanford parser)
Causality Relations
extraction
• Context inference
Thematic roles normalized
• Base forms
• URIs assignment
(Contextual Disambiguation)
Thematic roles assignment
• Based on VerbNet Index
State Inference
Causality Graph Building
• Built on 20 machines
• 300 million nodes
• 1 billion edges
• 13 million news articles in total
Outline
• Motivation
• Problem definition
• Solution
• Representation
• Algorithm
• Evaluation
Modeling an Event
Comparison between events (Canonical)
1. (Lexicon & Syntax) Language & wording
independent
2. (Semantic) Non ambiguous
Generalization / abstraction
Reasoning
Many philosophies
Property Exemplification of Events theory (Kim 1993)
Conceptual Dependency theory (Schank 1972)
Time
Event & Causality Representation
• Event Representation
• Causality Representation
5
Quantifier
kill
Action
Troops
Attribute
Afghan
1/2/1987
11:15AM +(3h)
Timeframe
Event2
“5 Afghan troops were killed”
Army
bombs
1/2/1987
11:00AM +(2h)
US Army
Timeframe
Action
US
Event1
Weapon
warehouse
Location
“US Army bombs a weapon
warehouse in Kabul with missiles”
Missiles
Kabul
Outline
• Motivation
• Problem definition
• Solution
• Representation
• Algorithm
• Causality Mining Process
• Evaluation
Machine Learning Problem Definition
Goal function:
Learning algorithm receives a set of examples
and produces a hypothesis
approximation of
which is good
Algorithm Outline
Learning Phase
1. Generalize events
2. Causality prediction rule generation
Prediction Phase
1. Finding similar generalized event
2. Application of causality prediction rule
Algorithm Outline
Learning Phase
1. Generalize events
1. How do we generalize objects?
2. How do we generalize actions?
3. How do we generalize an event?
2. Causality prediction rule generation
Generalizing Objects
the Russian Federation
Eastern Europe
RUS
Russian
Federation
185
Land border
Russia
Rouble (Rub)
643
USSR
China
Ontology – Linked data
Generalizing Actions
Levin classes (Levin 1993) – 270 classes
Class Hit-18.1
Roles and Restrictions: Agent[+int_control] Patient[+concrete]
Instrument[+concrete]
Members: bang, bash, hit, kick, ...
Frames:
Example
Syntax
Semantics
Name
Basic
Transitive
Paula hit the
ball
cause(Agent, E)manner(during(E),
directedmotion, Agent) !contact(during(E),
Agent V Patient
Agent, Patient) manner(end(E),forceful,
Agent) contact(end(E), Agent, Patient)
Generalizing Events:
Putting it all together
US
1/2/1987
11:00AM +(2h)
“NATO strikes an army base in
Baghdad”
NATO
Present
Event
Time-frame
strikes
Military
facility
bombs
Weapon
warehouse
City
Actor: [state of Nato]
Property: [Hit1.1]
Theme: [Military facility]
Location: [Arab City]
Action
US Army
Army base
Similar verb
Army
Baghdad
Generalization rule
Action
Country
Location
rdf:type
Time-frame
Past
Event
Instrument
1/2/1987
11:00AM +(2h)
Missiles
Location
Kabul
“US Army bombs a weapon
warehouse in Kabul with missiles”
Generalizing Events: HAC algorithm
Generalizing Events:
Event distance metric
US
1/2/1987
11:00AM +(2h)
“NATO strikes an army base in
Baghdad”
NATO
Present
Event
Time-frame
Action
Army
strikes
Military
facility
bombs
Weapon
warehouse
City
Action
US Army
Baghdad
Army base
Similar verb
Country
Location
rdf:type
Time-frame
Past
Event
Instrument
1/2/1987
11:00AM +(2h)
Missiles
Location
Kabul
“US Army bombs a weapon
warehouse in Kabul with missiles”
Algorithm Outline
Learning Phase
1. Generalize events
2. Causality prediction rule generation
Prediction Rule Generation
Time
5
Quantifier
kill
Action
Troops
Attribute
“5 Afghan troops were killed”
Afghan
1/2/1987
11:15AM +(3h)
Timeframe
Effect
Event
Nationality
Afghanistan
1/2/1987
11:00AM +(2h)
Type
Type
US
Army
bombs
US Army
Timeframe
Action
Country
Cause
Event
Weapon
warehouse
Location
“US Army bombs a weapon
warehouse in Kabul with missiles”
Missiles
Country
Kabul
EffectThemeAttribute =
CauseLocationCountry
Nationality
EffectAction=kill
EffectTheme=Troops
Algorithm Outline
Prediction Phase
1. Finding similar generalized event
2. Application of causality prediction rule
Finding Similar Generalized Event
0.2
“Baghdad bombing”
0.7
0.8
0.75
0.3
0.65
0.1
0.2
Prediction Rule Application
Time
kill
Action
Troops
Timeframe
T1 + ∆
Predicted
Effect
Event
bomb
T1
Timeframe
Nationality
Theme1
Country
Action
Actor1
Attribute
Input
Event
Location
Instrument1
Location1
EffectThemeAttribute =
CauseLocationCountry
Nationality
EffectAction=kill
EffectTheme=Troops
Outline
• Motivation
• Problem definition
• Solution
• Representation
• Algorithm
• Evaluation
Prediction Evaluation
Human Group 1:
• Mark events E that can cause other events.
Human Group 2:
• Given: Random sample of events from E , predictions and
time of events
• Search the web and give estimation on the prediction
accuracy
Prediction Accuracy Results
Highly certain
Certain
Algorithm
0.58
0.49
Humans
0.4
0.38
Causality Evaluation
Human Group 1:
• Mark events E for test for the second two control groups and the
algorithm.
Human Group 2:
• Given: Random sample of events from E.
• State what you think would happen following this event.
Human Group 3:
• Given: algorithm predictions + human (2nd group) predictions
• Evaluate the quality of the predictions
Causality Results
[0,1)
[1-2)
[2-3)
[3,4]
Avg. Rank
Avg. Accuracy
Algorithm 0
2
19
29
3.08
77%
Humans
3
24
23
2.86
72%
•
0
The results are statistically significant
Event
Predicted Event (Human)
Predicted event (Pundit)
Al-Qaida demands hostage Al-Qaida exchanges
exchange
hostage
A country will refuse the
demand
Volcano erupts in
Democratic Republic of
Congo
Scientists in Republic of
Congo investigate lava
beds
Thousands of people flee
from Congo
7.0 magnitude earthquake
strikes Haitian coast
Tsunami in Haiti effects
coast
Tsunami-warning is issued
2 Palestinians reportedly
shot dead by Israeli troops
Israeli citizens protest
against Palestinian leaders
War will be waged
Professor of Tehran
University killed in
bombing
Tehran students
remember slain professor
in memorial service
Professor funeral will be
held
Alleged drug kingpin
arrested in Mexico
Mafia kills people with
guns in town
Kingpin will be sent to
prison
UK bans Islamist group
Islamist group would
adopt another name in
the UK
Group will grow
China overtakes Germany German officials suspend
as world's biggest exporter tariffs
Wheat price will fall
Accuracy of Extraction
Extraction Evaluation
Action
Actor
Object
Instrument
Location Time
93%
74%
76%
79%
79%
Entity Ontology Matching
Actor
Object
Instrument
Location
84%
83%
79%
89%
100%
Related work
Causality Information Extraction
Goal: Extract causality relations from a text
Techniques:
1. Usage of handcrafted domain-specific patterns
[Kaplan and Berry-Rogghe, 1991]
2. Usage of handcrafted linguistic patterns
[Garcia 1997],[Khoo, Chan, &Niu 2000], [Girju &Moldovan 2002]
3. Semi-Supervised pattern learning approaches, based on text
features
[Blanco, Castell, &Moldovan 2008], [Sil & Huang & Yates 2010]
4. Supervised pattern learning approaches based on text
features
[Riloff 1996],[Riloff & Jones 1999], [Agichtein & Gravano, 2000; Lin & Pantel, 2001]
Related work
Temporal Information Extraction
Goal: Predicting the temporal order of events or time
expressions described in text
Technique: learn classifiers that predict a temporal order of a
pair of events based on a predefined features of the pair.
[Ling & Weld, 2010; Mani, Schiffman, & Zhang, 2003; Lapata & Lascarides,2006; Chambers, Wang, &
Jurafsky, 2007; Tatu & Srikanth, 2008; Yoshikawa, Riedel, Asahara, & Matsumoto, 2009]
Future work
• Going beyond human tagged examples
• Incorporating time into the equation
• When will correlation mean causality?
• Using other sources than news
• Incorporating real time data (Twitter, Facebook)
• Incorporating numerical data (Stocks, Weather, Forex)
• Can we predict general facts?
• Can a machine predict better than an expert?
Summary
• Canonical event representation
• Machine learning algorithm for events prediction
• Leveraging world knowledge for generalization
• Using text as human tagged examples
• Causality mining from text
• Contribution to machine common-sense
understanding
“The best way to predict the future is to invent it” [Alan Kay]