from narratology to formal narrative structures · pdf filefrom narratology to formal...
TRANSCRIPT
From Narratology to Formal Narrative Structures
What is Relevant in a Story?
Carlos León
University of Hamburg
Semantics and Syntax: A Legacy of Alan Turing
Isaac Newton Institute
Cambridge
12/1/2012
We use narratives every day
Tales
History
Law
Education
. . .
Narrative content is not only literature
A mathematical proof
What you did yesterday
Episodic memory
What you should do when you go to a restaurant
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 2 / 14
We use narratives every day
Tales
History
Law
Education
. . .
Narrative content is not only literature
A mathematical proof
What you did yesterday
Episodic memory
What you should do when you go to a restaurant
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 2 / 14
We use narratives every day
Tales
History
Law
Education
. . .
Narrative content is not only literature
A mathematical proof
What you did yesterday
Episodic memory
What you should do when you go to a restaurant
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 2 / 14
We use narratives every day
Tales
History
Law
Education
. . .
Narrative content is not only literature
A mathematical proof
What you did yesterday
Episodic memory
What you should do when you go to a restaurant
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 2 / 14
We use narratives every day
Tales
History
Law
Education
. . .
Narrative content is not only literature
A mathematical proof
What you did yesterday
Episodic memory
What you should do when you go to a restaurant
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 2 / 14
We use narratives every day
Tales
History
Law
Education
. . .
Narrative content is not only literature
A mathematical proof
What you did yesterday
Episodic memory
What you should do when you go to a restaurant
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 2 / 14
We use narratives every day
Tales
History
Law
Education
. . .
Narrative content is not only literature
A mathematical proof
What you did yesterday
Episodic memory
What you should do when you go to a restaurant
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 2 / 14
We use narratives every day
Tales
History
Law
Education
. . .
Narrative content is not only literature
A mathematical proof
What you did yesterday
Episodic memory
What you should do when you go to a restaurant
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 2 / 14
We use narratives every day
Tales
History
Law
Education
. . .
Narrative content is not only literature
A mathematical proof
What you did yesterday
Episodic memory
What you should do when you go to a restaurant
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 2 / 14
We use narratives every day
Tales
History
Law
Education
. . .
Narrative content is not only literature
A mathematical proof
What you did yesterday
Episodic memory
What you should do when you go to a restaurant
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 2 / 14
We use narratives every day
Tales
History
Law
Education
. . .
Narrative content is not only literature
A mathematical proof
What you did yesterday
Episodic memory
What you should do when you go to a restaurant
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 2 / 14
Computational Narrative is important in AI
Narratives can store a huge amount of common sense knowledge
This is relevant for computation in general and AI in particular
From a computational point of view, knowledge is not only representation
Knowledge management is also a crucial part (time, events, relations, causality)
General common sense knowledge can not be represented and processed without
taking narrative properties into account
We are interested in the properties that define narrative similarity
If a mathematical proof is a narrative, we could find out whether two proofs are basically the same
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 3 / 14
Computational Narrative is important in AI
Narratives can store a huge amount of common sense knowledge
This is relevant for computation in general and AI in particular
From a computational point of view, knowledge is not only representation
Knowledge management is also a crucial part (time, events, relations, causality)
General common sense knowledge can not be represented and processed without
taking narrative properties into account
We are interested in the properties that define narrative similarity
If a mathematical proof is a narrative, we could find out whether two proofs are basically the same
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 3 / 14
Computational Narrative is important in AI
Narratives can store a huge amount of common sense knowledge
This is relevant for computation in general and AI in particular
From a computational point of view, knowledge is not only representation
Knowledge management is also a crucial part (time, events, relations, causality)
General common sense knowledge can not be represented and processed without
taking narrative properties into account
We are interested in the properties that define narrative similarity
If a mathematical proof is a narrative, we could find out whether two proofs are basically the same
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 3 / 14
Computational Narrative is important in AI
Narratives can store a huge amount of common sense knowledge
This is relevant for computation in general and AI in particular
From a computational point of view, knowledge is not only representation
Knowledge management is also a crucial part (time, events, relations, causality)
General common sense knowledge can not be represented and processed without
taking narrative properties into account
We are interested in the properties that define narrative similarity
If a mathematical proof is a narrative, we could find out whether two proofs are basically the same
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 3 / 14
Computational Narrative is important in AI
Narratives can store a huge amount of common sense knowledge
This is relevant for computation in general and AI in particular
From a computational point of view, knowledge is not only representation
Knowledge management is also a crucial part (time, events, relations, causality)
General common sense knowledge can not be represented and processed without
taking narrative properties into account
We are interested in the properties that define narrative similarity
If a mathematical proof is a narrative, we could find out whether two proofs are basically the same
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 3 / 14
Computational Narrative is important in AI
Narratives can store a huge amount of common sense knowledge
This is relevant for computation in general and AI in particular
From a computational point of view, knowledge is not only representation
Knowledge management is also a crucial part (time, events, relations, causality)
General common sense knowledge can not be represented and processed without
taking narrative properties into account
We are interested in the properties that define narrative similarity
If a mathematical proof is a narrative, we could find out whether two proofs are basically the same
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 3 / 14
Computational Narrative is important in AI
Narratives can store a huge amount of common sense knowledge
This is relevant for computation in general and AI in particular
From a computational point of view, knowledge is not only representation
Knowledge management is also a crucial part (time, events, relations, causality)
General common sense knowledge can not be represented and processed without
taking narrative properties into account
We are interested in the properties that define narrative similarity
If a mathematical proof is a narrative, we could find out whether two proofs are basically the same
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 3 / 14
Computational Narrative is important in AI
Narratives can store a huge amount of common sense knowledge
This is relevant for computation in general and AI in particular
From a computational point of view, knowledge is not only representation
Knowledge management is also a crucial part (time, events, relations, causality)
General common sense knowledge can not be represented and processed without
taking narrative properties into account
We are interested in the properties that define narrative similarity
If a mathematical proof is a narrative, we could find out whether two proofs are basically the same
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 3 / 14
Narratology influences Computational Narrative
Classic Narratology studies the structural properties of stories
Beginning, middle, end
Exposition, climax, denouement
This suggest some sense of syntax in stories
Can we properly define a narration in terms of its structure?
Can we properly represent a story computationally in terms of its structure?
Narratology assumes cognitive properties way beyond AI
How do we know if a narrative conflict has been solved?
How does a computer know if a narrative conflict has been solved?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 4 / 14
Narratology influences Computational Narrative
Classic Narratology studies the structural properties of stories
Beginning, middle, end
Exposition, climax, denouement
This suggest some sense of syntax in stories
Can we properly define a narration in terms of its structure?
Can we properly represent a story computationally in terms of its structure?
Narratology assumes cognitive properties way beyond AI
How do we know if a narrative conflict has been solved?
How does a computer know if a narrative conflict has been solved?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 4 / 14
Narratology influences Computational Narrative
Classic Narratology studies the structural properties of stories
Beginning, middle, end
Exposition, climax, denouement
This suggest some sense of syntax in stories
Can we properly define a narration in terms of its structure?
Can we properly represent a story computationally in terms of its structure?
Narratology assumes cognitive properties way beyond AI
How do we know if a narrative conflict has been solved?
How does a computer know if a narrative conflict has been solved?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 4 / 14
Narratology influences Computational Narrative
Classic Narratology studies the structural properties of stories
Beginning, middle, end
Exposition, climax, denouement
This suggest some sense of syntax in stories
Can we properly define a narration in terms of its structure?
Can we properly represent a story computationally in terms of its structure?
Narratology assumes cognitive properties way beyond AI
How do we know if a narrative conflict has been solved?
How does a computer know if a narrative conflict has been solved?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 4 / 14
Narratology influences Computational Narrative
Classic Narratology studies the structural properties of stories
Beginning, middle, end
Exposition, climax, denouement
This suggest some sense of syntax in stories
Can we properly define a narration in terms of its structure?
Can we properly represent a story computationally in terms of its structure?
Narratology assumes cognitive properties way beyond AI
How do we know if a narrative conflict has been solved?
How does a computer know if a narrative conflict has been solved?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 4 / 14
Narratology influences Computational Narrative
Classic Narratology studies the structural properties of stories
Beginning, middle, end
Exposition, climax, denouement
This suggest some sense of syntax in stories
Can we properly define a narration in terms of its structure?
Can we properly represent a story computationally in terms of its structure?
Narratology assumes cognitive properties way beyond AI
How do we know if a narrative conflict has been solved?
How does a computer know if a narrative conflict has been solved?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 4 / 14
Narratology influences Computational Narrative
Classic Narratology studies the structural properties of stories
Beginning, middle, end
Exposition, climax, denouement
This suggest some sense of syntax in stories
Can we properly define a narration in terms of its structure?
Can we properly represent a story computationally in terms of its structure?
Narratology assumes cognitive properties way beyond AI
How do we know if a narrative conflict has been solved?
How does a computer know if a narrative conflict has been solved?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 4 / 14
Narratology influences Computational Narrative
Classic Narratology studies the structural properties of stories
Beginning, middle, end
Exposition, climax, denouement
This suggest some sense of syntax in stories
Can we properly define a narration in terms of its structure?
Can we properly represent a story computationally in terms of its structure?
Narratology assumes cognitive properties way beyond AI
How do we know if a narrative conflict has been solved?
How does a computer know if a narrative conflict has been solved?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 4 / 14
Narratology influences Computational Narrative
Classic Narratology studies the structural properties of stories
Beginning, middle, end
Exposition, climax, denouement
This suggest some sense of syntax in stories
Can we properly define a narration in terms of its structure?
Can we properly represent a story computationally in terms of its structure?
Narratology assumes cognitive properties way beyond AI
How do we know if a narrative conflict has been solved?
How does a computer know if a narrative conflict has been solved?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 4 / 14
Narratology influences Computational Narrative
Classic Narratology studies the structural properties of stories
Beginning, middle, end
Exposition, climax, denouement
This suggest some sense of syntax in stories
Can we properly define a narration in terms of its structure?
Can we properly represent a story computationally in terms of its structure?
Narratology assumes cognitive properties way beyond AI
How do we know if a narrative conflict has been solved?
How does a computer know if a narrative conflict has been solved?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 4 / 14
A computational definition of syntax
When the rules for using all the symbols are known by the computer, no external human
understanding is needed
Computers are great at applying rules to symbols
Therefore, we’d like to have a syntactical definition of narratives
A useful structural definition of stories is maybe less natural than focusing on the chain of
events, but can be more powerful as a computational solution
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 5 / 14
A computational definition of syntax
When the rules for using all the symbols are known by the computer, no external human
understanding is needed
Computers are great at applying rules to symbols
Therefore, we’d like to have a syntactical definition of narratives
A useful structural definition of stories is maybe less natural than focusing on the chain of
events, but can be more powerful as a computational solution
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 5 / 14
A computational definition of syntax
When the rules for using all the symbols are known by the computer, no external human
understanding is needed
Computers are great at applying rules to symbols
Therefore, we’d like to have a syntactical definition of narratives
A useful structural definition of stories is maybe less natural than focusing on the chain of
events, but can be more powerful as a computational solution
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 5 / 14
A computational definition of syntax
When the rules for using all the symbols are known by the computer, no external human
understanding is needed
Computers are great at applying rules to symbols
Therefore, we’d like to have a syntactical definition of narratives
A useful structural definition of stories is maybe less natural than focusing on the chain of
events, but can be more powerful as a computational solution
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 5 / 14
A computational definition of syntax
When the rules for using all the symbols are known by the computer, no external human
understanding is needed
Computers are great at applying rules to symbols
Therefore, we’d like to have a syntactical definition of narratives
A useful structural definition of stories is maybe less natural than focusing on the chain of
events, but can be more powerful as a computational solution
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 5 / 14
A computational definition of semantics
A story (as a computer sees it)
A, B
Can we assign any meaning to these symbols?
happy(john), give(john, mary, ball)
die(john), give(john, mary, ball)
die(cat), eat(cat, fish)
What the symbols mean to us (semantics) is crucial
What amount of semantics do we need?
To what extent can we rely on structural properties?
The more developed the computational system, the more complicated the symbols
that can be processed. Do these complex symbols become syntax?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 6 / 14
A computational definition of semantics
A story (as a computer sees it)
A, B
Can we assign any meaning to these symbols?
happy(john), give(john, mary, ball)
die(john), give(john, mary, ball)
die(cat), eat(cat, fish)
What the symbols mean to us (semantics) is crucial
What amount of semantics do we need?
To what extent can we rely on structural properties?
The more developed the computational system, the more complicated the symbols
that can be processed. Do these complex symbols become syntax?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 6 / 14
A computational definition of semantics
A story (as a computer sees it)
A, B
Can we assign any meaning to these symbols?
happy(john), give(john, mary, ball)
die(john), give(john, mary, ball)
die(cat), eat(cat, fish)
What the symbols mean to us (semantics) is crucial
What amount of semantics do we need?
To what extent can we rely on structural properties?
The more developed the computational system, the more complicated the symbols
that can be processed. Do these complex symbols become syntax?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 6 / 14
A computational definition of semantics
A story (as a computer sees it)
A, B
Can we assign any meaning to these symbols?
happy(john), give(john, mary, ball)
die(john), give(john, mary, ball)
die(cat), eat(cat, fish)
What the symbols mean to us (semantics) is crucial
What amount of semantics do we need?
To what extent can we rely on structural properties?
The more developed the computational system, the more complicated the symbols
that can be processed. Do these complex symbols become syntax?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 6 / 14
A computational definition of semantics
A story (as a computer sees it)
A, B
Can we assign any meaning to these symbols?
happy(john), give(john, mary, ball)
die(john), give(john, mary, ball)
die(cat), eat(cat, fish)
What the symbols mean to us (semantics) is crucial
What amount of semantics do we need?
To what extent can we rely on structural properties?
The more developed the computational system, the more complicated the symbols
that can be processed. Do these complex symbols become syntax?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 6 / 14
A computational definition of semantics
A story (as a computer sees it)
A, B
Can we assign any meaning to these symbols?
happy(john), give(john, mary, ball)
die(john), give(john, mary, ball)
die(cat), eat(cat, fish)
What the symbols mean to us (semantics) is crucial
What amount of semantics do we need?
To what extent can we rely on structural properties?
The more developed the computational system, the more complicated the symbols
that can be processed. Do these complex symbols become syntax?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 6 / 14
A computational definition of semantics
A story (as a computer sees it)
A, B
Can we assign any meaning to these symbols?
happy(john), give(john, mary, ball)
die(john), give(john, mary, ball)
die(cat), eat(cat, fish)
What the symbols mean to us (semantics) is crucial
What amount of semantics do we need?
To what extent can we rely on structural properties?
The more developed the computational system, the more complicated the symbols
that can be processed. Do these complex symbols become syntax?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 6 / 14
A computational definition of semantics
A story (as a computer sees it)
A, B
Can we assign any meaning to these symbols?
happy(john), give(john, mary, ball)
die(john), give(john, mary, ball)
die(cat), eat(cat, fish)
What the symbols mean to us (semantics) is crucial
What amount of semantics do we need?
To what extent can we rely on structural properties?
The more developed the computational system, the more complicated the symbols
that can be processed. Do these complex symbols become syntax?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 6 / 14
A computational definition of semantics
A story (as a computer sees it)
A, B
Can we assign any meaning to these symbols?
happy(john), give(john, mary, ball)
die(john), give(john, mary, ball)
die(cat), eat(cat, fish)
What the symbols mean to us (semantics) is crucial
What amount of semantics do we need?
To what extent can we rely on structural properties?
The more developed the computational system, the more complicated the symbols
that can be processed. Do these complex symbols become syntax?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 6 / 14
A computational definition of semantics
A story (as a computer sees it)
A, B
Can we assign any meaning to these symbols?
happy(john), give(john, mary, ball)
die(john), give(john, mary, ball)
die(cat), eat(cat, fish)
What the symbols mean to us (semantics) is crucial
What amount of semantics do we need?
To what extent can we rely on structural properties?
The more developed the computational system, the more complicated the symbols
that can be processed. Do these complex symbols become syntax?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 6 / 14
A computational definition of semantics
A story (as a computer sees it)
A, B
Can we assign any meaning to these symbols?
happy(john), give(john, mary, ball)
die(john), give(john, mary, ball)
die(cat), eat(cat, fish)
What the symbols mean to us (semantics) is crucial
What amount of semantics do we need?
To what extent can we rely on structural properties?
The more developed the computational system, the more complicated the symbols
that can be processed. Do these complex symbols become syntax?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 6 / 14
Many Different Formal Systems
Knowledge must be represented in terms of manageable symbols
A, B, C, D, E, · · ·
a
c
b
e
d
hate(john, mary), attack(john, mary), die(mary), happy(john)
Which one is better?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 7 / 14
Many Different Formal Systems
Knowledge must be represented in terms of manageable symbols
A, B, C, D, E, · · ·
a
c
b
e
d
hate(john, mary), attack(john, mary), die(mary), happy(john)
Which one is better?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 7 / 14
Many Different Formal Systems
Knowledge must be represented in terms of manageable symbols
A, B, C, D, E, · · ·
a
c
b
e
d
hate(john, mary), attack(john, mary), die(mary), happy(john)
Which one is better?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 7 / 14
Many Different Formal Systems
Knowledge must be represented in terms of manageable symbols
A, B, C, D, E, · · ·
a
c
b
e
d
hate(john, mary), attack(john, mary), die(mary), happy(john)
Which one is better?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 7 / 14
Many Different Formal Systems
Knowledge must be represented in terms of manageable symbols
A, B, C, D, E, · · ·
a
c
b
e
d
hate(john, mary), attack(john, mary), die(mary), happy(john)
Which one is better?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 7 / 14
Many Different Formal Systems
Knowledge must be represented in terms of manageable symbols
A, B, C, D, E, · · ·
a
c
b
e
d
hate(john, mary), attack(john, mary), die(mary), happy(john)
Which one is better?
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 7 / 14
We do not know which formalization is the correct one
When identifying narrative similarity, the level of abstraction is important
Plot
Sequence of events
Text
. . .
We are trying to find out what the correct formal system is
A family of formal systems with different levels of granularity is probably the best solution
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 8 / 14
We do not know which formalization is the correct one
When identifying narrative similarity, the level of abstraction is important
Plot
Sequence of events
Text
. . .
We are trying to find out what the correct formal system is
A family of formal systems with different levels of granularity is probably the best solution
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 8 / 14
We do not know which formalization is the correct one
When identifying narrative similarity, the level of abstraction is important
Plot
Sequence of events
Text
. . .
We are trying to find out what the correct formal system is
A family of formal systems with different levels of granularity is probably the best solution
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 8 / 14
We do not know which formalization is the correct one
When identifying narrative similarity, the level of abstraction is important
Plot
Sequence of events
Text
. . .
We are trying to find out what the correct formal system is
A family of formal systems with different levels of granularity is probably the best solution
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 8 / 14
We do not know which formalization is the correct one
When identifying narrative similarity, the level of abstraction is important
Plot
Sequence of events
Text
. . .
We are trying to find out what the correct formal system is
A family of formal systems with different levels of granularity is probably the best solution
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 8 / 14
We do not know which formalization is the correct one
When identifying narrative similarity, the level of abstraction is important
Plot
Sequence of events
Text
. . .
We are trying to find out what the correct formal system is
A family of formal systems with different levels of granularity is probably the best solution
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 8 / 14
We do not know which formalization is the correct one
When identifying narrative similarity, the level of abstraction is important
Plot
Sequence of events
Text
. . .
We are trying to find out what the correct formal system is
A family of formal systems with different levels of granularity is probably the best solution
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 8 / 14
We do not know which formalization is the correct one
When identifying narrative similarity, the level of abstraction is important
Plot
Sequence of events
Text
. . .
We are trying to find out what the correct formal system is
A family of formal systems with different levels of granularity is probably the best solution
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 8 / 14
What to include in the formalization?
The formalization process implies abstraction
This abstraction usually implies some sense of summarization
Mary was a woman. Mary hated John. Mary attacked John. John died. Mary was
happy.
John was a man. John hated Mary. John attacked Mary. Mary died. John was happy.
Not all the information is relevant at a certain level
It is important to finding out what levels of similarity we use
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 9 / 14
What to include in the formalization?
The formalization process implies abstraction
This abstraction usually implies some sense of summarization
Mary was a woman. Mary hated John. Mary attacked John. John died. Mary was
happy.
John was a man. John hated Mary. John attacked Mary. Mary died. John was happy.
Not all the information is relevant at a certain level
It is important to finding out what levels of similarity we use
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 9 / 14
What to include in the formalization?
The formalization process implies abstraction
This abstraction usually implies some sense of summarization
Mary was a woman. Mary hated John. Mary attacked John. John died. Mary was
happy.
John was a man. John hated Mary. John attacked Mary. Mary died. John was happy.
Not all the information is relevant at a certain level
It is important to finding out what levels of similarity we use
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 9 / 14
What to include in the formalization?
The formalization process implies abstraction
This abstraction usually implies some sense of summarization
Mary was a woman. Mary hated John. Mary attacked John. John died. Mary was
happy.
John was a man. John hated Mary. John attacked Mary. Mary died. John was happy.
Not all the information is relevant at a certain level
It is important to finding out what levels of similarity we use
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 9 / 14
What to include in the formalization?
The formalization process implies abstraction
This abstraction usually implies some sense of summarization
Mary was a woman. Mary hated John. Mary attacked John. John died. Mary was
happy.
John was a man. John hated Mary. John attacked Mary. Mary died. John was happy.
Not all the information is relevant at a certain level
It is important to finding out what levels of similarity we use
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 9 / 14
What to include in the formalization?
The formalization process implies abstraction
This abstraction usually implies some sense of summarization
Mary was a woman. Mary hated John. Mary attacked John. John died. Mary was
happy.
John was a man. John hated Mary. John attacked Mary. Mary died. John was happy.
Not all the information is relevant at a certain level
It is important to finding out what levels of similarity we use
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 9 / 14
What to include in the formalization?
The formalization process implies abstraction
This abstraction usually implies some sense of summarization
Mary was a woman. Mary hated John. Mary attacked John. John died. Mary was
happy.
John was a man. John hated Mary. John attacked Mary. Mary died. John was happy.
Not all the information is relevant at a certain level
It is important to finding out what levels of similarity we use
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 9 / 14
How do people formalize?
Stories and everyday narrative is not formally represented
Therefore, a formal system is not enough, a formalizing process is needed
In order to be able to compare the structural properties, we need a robust formalizing
methodology
Computers can not do this yet, humans must perform formalization
A natural formalization does not exist
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 10 / 14
How do people formalize?
Stories and everyday narrative is not formally represented
Therefore, a formal system is not enough, a formalizing process is needed
In order to be able to compare the structural properties, we need a robust formalizing
methodology
Computers can not do this yet, humans must perform formalization
A natural formalization does not exist
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 10 / 14
How do people formalize?
Stories and everyday narrative is not formally represented
Therefore, a formal system is not enough, a formalizing process is needed
In order to be able to compare the structural properties, we need a robust formalizing
methodology
Computers can not do this yet, humans must perform formalization
A natural formalization does not exist
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 10 / 14
How do people formalize?
Stories and everyday narrative is not formally represented
Therefore, a formal system is not enough, a formalizing process is needed
In order to be able to compare the structural properties, we need a robust formalizing
methodology
Computers can not do this yet, humans must perform formalization
A natural formalization does not exist
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 10 / 14
How do people formalize?
Stories and everyday narrative is not formally represented
Therefore, a formal system is not enough, a formalizing process is needed
In order to be able to compare the structural properties, we need a robust formalizing
methodology
Computers can not do this yet, humans must perform formalization
A natural formalization does not exist
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 10 / 14
Is everyone well trained?
Formalizing narratives is not intuitive
People must be trained
Therefore only experts can formalize a story
We need control over the formalization process
Without a natural formalization, the comparison of stories is tainted unless the
formalization is controlled very carefully
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 11 / 14
Is everyone well trained?
Formalizing narratives is not intuitive
People must be trained
Therefore only experts can formalize a story
We need control over the formalization process
Without a natural formalization, the comparison of stories is tainted unless the
formalization is controlled very carefully
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 11 / 14
Is everyone well trained?
Formalizing narratives is not intuitive
People must be trained
Therefore only experts can formalize a story
We need control over the formalization process
Without a natural formalization, the comparison of stories is tainted unless the
formalization is controlled very carefully
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 11 / 14
Is everyone well trained?
Formalizing narratives is not intuitive
People must be trained
Therefore only experts can formalize a story
We need control over the formalization process
Without a natural formalization, the comparison of stories is tainted unless the
formalization is controlled very carefully
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 11 / 14
Is everyone well trained?
Formalizing narratives is not intuitive
People must be trained
Therefore only experts can formalize a story
We need control over the formalization process
Without a natural formalization, the comparison of stories is tainted unless the
formalization is controlled very carefully
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 11 / 14
Is everyone well trained?
Formalizing narratives is not intuitive
People must be trained
Therefore only experts can formalize a story
We need control over the formalization process
Without a natural formalization, the comparison of stories is tainted unless the
formalization is controlled very carefully
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 11 / 14
Experiments for finding out how humans create formalizations
Propp
Narrative annotation with proppian functions
Results show that do not agree with Propp
Basic summarization
We want to know which parts are highlighted by test subjects.
Simple sentences to lesen linguistic side-effects
We plan to compare the results against proppian experiment
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 12 / 14
Experiments for finding out how humans create formalizations
Propp
Narrative annotation with proppian functions
Results show that do not agree with Propp
Basic summarization
We want to know which parts are highlighted by test subjects.
Simple sentences to lesen linguistic side-effects
We plan to compare the results against proppian experiment
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 12 / 14
Experiments for finding out how humans create formalizations
Propp
Narrative annotation with proppian functions
Results show that do not agree with Propp
Basic summarization
We want to know which parts are highlighted by test subjects.
Simple sentences to lesen linguistic side-effects
We plan to compare the results against proppian experiment
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 12 / 14
Experiments for finding out how humans create formalizations
Propp
Narrative annotation with proppian functions
Results show that do not agree with Propp
Basic summarization
We want to know which parts are highlighted by test subjects.
Simple sentences to lesen linguistic side-effects
We plan to compare the results against proppian experiment
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 12 / 14
Experiments for finding out how humans create formalizations
Propp
Narrative annotation with proppian functions
Results show that do not agree with Propp
Basic summarization
We want to know which parts are highlighted by test subjects.
Simple sentences to lesen linguistic side-effects
We plan to compare the results against proppian experiment
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 12 / 14
Experiments for finding out how humans create formalizations
Propp
Narrative annotation with proppian functions
Results show that do not agree with Propp
Basic summarization
We want to know which parts are highlighted by test subjects.
Simple sentences to lesen linguistic side-effects
We plan to compare the results against proppian experiment
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 12 / 14
Experiments for finding out how humans create formalizations
Propp
Narrative annotation with proppian functions
Results show that do not agree with Propp
Basic summarization
We want to know which parts are highlighted by test subjects.
Simple sentences to lesen linguistic side-effects
We plan to compare the results against proppian experiment
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 12 / 14
Experiments for finding out how humans create formalizations
Propp
Narrative annotation with proppian functions
Results show that do not agree with Propp
Basic summarization
We want to know which parts are highlighted by test subjects.
Simple sentences to lesen linguistic side-effects
We plan to compare the results against proppian experiment
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 12 / 14
Conclusions
Narratives represent an important part of human knowledge
Narratology offers structural analysis of narratives
From a formal point of view, the threshold between semantics and syntax is not clear
A robust formal system for story similarity must have different levels of granularity
Some levels will be more syntactical
Some levels will be more semantical
The formalization task is complex and must be controlled carefully
Experiments on human behavior (tagging, summarization, formalization) can help to
understand what the relevant parts of a story are
C. León (U. Hamburg) Formal Narrative Structures 12/1/2012 13 / 14
From Narratology to Formal Narrative Structures
What is Relevant in a Story?
Carlos León
University of Hamburg
Semantics and Syntax: A Legacy of Alan Turing
Isaac Newton Institute
Cambridge
12/1/2012