legal knowledge conveyed by narratives: towards a representational model
TRANSCRIPT
Legal Knowledge Conveyed by Narratives
2 August 2014 - CMN presentation
Giovanni Sileno [email protected] Alexander BoerTom van Engers
Leibniz Center for Law – University of Amsterdam
Towards a Representational Model
The social function of “stories”
Stories are “constituents of human memory, knowledge, and social communication”Schank and Abelson [1995], Knowledge and memory: the real story
The social function of “stories”
“Many different root metaphors have been put forth to represent the essential nature of human beings: homo faber, homo economous, homo politicus, [...], rational man. I now propose homo narrans to be added to the list.
Fischer [1984], Narration as a Human Communication Paradigm
What “stories” are:Not only fictional narrations.. but also:
• personal experiences• journalistic reports • medical cases• legal cases• …
What is a legal case about?witnesses’, experts’ speeches used to justify a certain story reconstruction
• events occurred in a social scenariolawyers’, judges’ speeches used to justify a certain legal interpretation of this story
• legal implications
What is a legal case about?witnesses’, experts’ speeches used to justify a certain story reconstruction
• events occurred in a social scenariolawyers’, judges’ speeches used to justify a certain legal interpretation of the story
• legal implications
What is a legal case about?witnesses’, experts’ speeches used to justify a certain story reconstruction
• events occurred in a social scenariolawyers’, judges’ speeches used to justify a certain legal interpretation of the story
• legal implications
• nested narratives/speech acts• originated from a failure of norm. expectations• informative intent within the legal system• brings part of background to the foreground
A relevant issue
Unfortunately, stories, by their own nature, are partial (ill-defined) representations.
What is in a story depends on • default assumptions • common-sense knowledge • expertise • interest• focus • ...
A relevant issue
Unfortunately, stories, by their own nature, are partial (ill-defined) representations.
What is told in a story depends on • default assumptions • common-sense knowledge • expertise • interest• focus • intent...
of the narrator !
A relevant issue
Unfortunately, stories, by their own nature, are partial (ill-defined) representations.
What is read in a story depends on • default assumptions • common-sense knowledge • expertise • interest• focus • intent...
of the listener !
Three ontological domains
• discourse: the signal• story: the meaning • conversation: the context, i.e. the
knowledge and intent of narrator/listener and those ascribed to the other party
Three ontological domains
• discourse: the signal• story: the meaning • conversation: the context, i.e. the
knowledge and intent of narrator/listener and those ascribed to the other party
narrative interpretation:discourse + conversation story
narrative generation:conversation + story discourse
Pierson v Post
Post was hunting a fox with a horse and hounds in a wild and uninhabited land, and was about to catch it, but Pierson, although conscious of Post's pursuit, intercepted the fox, killed it and took the animal.
Tompkins: Possession of a fera naturae occurs only if there is occupancy, i.e. taking physical possession. Pierson took it, so he owns it.
Pierson v Post
Post was hunting a fox with a horse and hounds in a wild and uninhabited land, and was about to catch it, but Pierson, although conscious of Post's pursuit, intercepted the fox, killed it and took the animal.
Livingston: If someone starts and hunts a fox with hounds in a uninhabited land has a right of taking the fox on any other person who saw he was pursuing it.
“Shallow” story model
• sequence of events
Post was hunting a fox with a horse and hounds in a wild and uninhabited land, and was about to catch it, but Pierson, although conscious of Post's pursuit, intercepted the fox, killed it and took the animal. [..]
• sequence of events• occurring at certain circumstances
Post was hunting a fox with a horse and hounds in a wild and uninhabited land, and was about to catch it, but Pierson, although conscious of Post's pursuit, intercepted the fox, killed it and took the animal. [..]
“Shallow” story model
Post was hunting a fox with a horse and hounds in a wild and uninhabited land, and was about to catch it, but Pierson, although conscious of Post's pursuit, intercepted the fox, killed it and took the animal.
Possession of a fera naturae occurs only if there is occupancy, i.e. taking physical possession. Pierson took it, so he owns it.
• sequence of events• occurring at certain circumstances+ explicit mechanisms relating
events/circumstances
“Shallow” story model
“Shallow” story model
• a sequence of events• occurring at certain circumstances+ explicit mechanisms relating events/
circumstances
• What about the implicit mechanisms?
Our objective
We look for a methodology to
• acquire in a computational form
• the systematic knowledge
• concerning a social scenario,
• presented through a narrative (e.g. a legal case, a scenario given by an expert, etc.)
• allowing alternative interpretations
Requirements
• bypass natural language issues• we are not targeting narrative
comprehension, but scenario acquisition from different interpreters
• target non-IT experts (in principle) we refer mostly to diagrams, or programming based
on high level and “intuitive” languages affinity with scenario-based modeling
Requirements
From “Shallow” to “Deep” story model
Issues:• consecutiveness vs consequence
“the mainspring of the narrative activity is to be traced to that very confusion between consecutiveness and consequence, what-comes-after being read in a narrative as what-is-caused-by”, Barthes [1968]
• story-relative vs discourse-relative timelinesordering as events occur vs how they are told/observed
Three levels of constraints• weak: discourse ordering • medium: relative/absolute time indexing • strong: dependencies (logic or causal)
From “Shallow” to “Deep” story model
Three levels of constraints• weak: discourse ordering • medium: relative/absolute time indexing • strong: dependencies (logic or causal)
The last is by far the most important to our scope: problem of Contingency vs Contextuality
From “Shallow” to “Deep” story model
Three levels of constraints• weak: discourse ordering • medium: relative/absolute time indexing • strong: dependencies (logic or causal)
The last is by far the most important to our scope: problem of Contingency vs Contextuality
Shallow
Deep
From “Shallow” to “Deep” story model
An important step is the recognition of sub-systems operating concurrently (e.g. agents, concurrent cognitive modules).
What is necessarily said in a sequential way, may in fact be simultaneous.
From “Shallow” to “Deep” story model
An important step is the recognition of sub-systems operating concurrently (e.g. agents, concurrent cognitive modules).
What is necessarily said in a sequential way, may in fact be simultaneous.
A specific story provides a synchronization between concurrent systems (as agents).
From “Shallow” to “Deep” story model
Agent story scheme
Motive Intent Action ActBex and Verheij (2011), Pennington and Hastie (1993)
• the scheme is used by investigators as template to map plausible scenarios, and to anchor evidence.
• intent is meaningful for legal purposes (both in design that adjudication)
• the existence or absence of motive may also influence the jury
Agent story scheme (Post)
Motive Intent Action Act (Failure)
Motive
Post intends to hunt that fox
Intent
Post sees a fox
Agent story scheme (Post)
Motive Intent Action Act (Failure)
Motive
Post intends to hunt that fox
Intent
Post sees a fox
1. Post sees a fox.
Agent story scheme (Post)
Motive Intent Action Act (Failure)
1. Post sees a fox.2. Post intends to
hunt that fox.
Is this enough?
Motive
Post intends to hunt that fox
Intent
Post sees a fox
Agent story scheme (Post)
Motive Intent Action Act (Failure)
• Considering only these events the model is not enough specified.
We need circumstantial conditions
Motive Intent Action Act (Failure)
Post is hunting foxes
Agent story scheme (Post)
Motivation
+ Motivation
Motive
Post intends to hunt that fox
Intent
Post sees a fox
Agent story scheme (Post)
Motive Intent Action Act (Failure)+ Motivation
Post is hunting foxesMotivation
Motive
Post intends to hunt that fox
Intent
Post sees a fox
Agent story scheme (Post)
Motive Intent Action Act (Failure)+ Motivation
Post is hunting foxesMotivation
Motive
Post intends to hunt that fox
Intent
Post sees a fox
Agent story scheme (Post)
Motive Intent Action Act (Failure)
Intent
Post is hunting the fox
Action
+ Affordance
Post thinks he has the power to hunt that fox
Post intends to hunt that fox
Affordance
Agent story scheme (Post)
Motive Intent Action Act (Failure)+ Affordance
Post is hunting the fox
Post thinks he has the power to hunt that fox
Post intends to hunt that fox
Intent
Action
Affordance
Motive Intent Action Act (Failure)
Action
Post has hunted the fox
Act
+ Disposition
Post has actually the power to hunt that fox
Agent story scheme (Post)
Disposition
Post is hunting..
Agent story scheme (Pierson)
Motive Intent Action Act (Failure)+ Disposition
Pierson has hunted the fox
Pierson has actually the power to hunt that fox
Action
Act
Pierson is hunting..
Disposition
Agent story scheme (Pierson)
Motive Intent Action Act (Failure)+ Disposition
Pierson has hunted the fox
Pierson has actually the power to hunt that fox
Action
Act
Pierson is hunting..
Disposition
Agent story scheme (Pierson)
Motive Intent Action Act (Failure)+ Disposition
Pierson has hunted the fox
Pierson has actually the power to hunt that fox
Action
Act
Pierson is hunting..
Disposition
Failures and social failures
• The difference between expectations and actual outcome allows to include in the model the computation of failures (cf. Plot units)
useful for validation!
Failures
Action failure
Post is hunting the foxPierson hashunted the fox= Post has nothunted the fox
Post failed to hunt the fox
Action
The explicit/implicit expectations (intentions and actions) allow to compute failures
Failures
Pierson hashunted the fox= Post has nothunted the fox
Action failure
Post is hunting the fox
Post failed to hunt the fox
Action
The explicit/implicit expectations (intentions and actions) allow to compute failures
Failures
Pierson hashunted the fox= Post has nothunted the fox
Action failure
Post is hunting the fox
Post failed to hunt the fox
Action
The explicit/implicit expectations (intentions and actions) allow to compute failures
Social failures
Social failure
Social expectation
Pierson is not permitted to hunt the fox
Pierson has hunted the fox
Normative expectations (obligations, permissions and institutional powers) allow to determine social failures
Social failures
Social failure
Social expectation
Pierson is not permitted to hunt the fox
Pierson has hunted the fox
Normative expectations (obligations, permissions and institutional powers) allow to determine social failures
Social failures
Social failure
Social expectation
Pierson is not permitted to hunt the fox
Pierson has hunted the fox
Normative expectations (obligations, permissions and institutional powers) allow to determine social failures
Discussion
• This contribution is based on a weak definition of validity for story models
• As long as the execution of the given mechanisms with the right synchronization produces the narrated events, the model is valid.
allow for alternative interpretations/fabulae
Discussion
• We do not consider the problem of increased depth as the most difficult issue in our domain
1. The explicit modeling of the scheme of a case is useful for clarification purposes
Discussion
• We do not consider the problem of increased depth as the most difficult issue in our domain
2. Some story components may be easily reused
Discussion
• We do not consider the problem of increased depth as the most difficult issue in our domain
2. Some story components may be easily reused
3. If they cannot be reused but seems applicable, the system can ask the modeler to provide circumstantial distinction
constructivist acquisition model
Discussion
• We do not consider the problem of increased depth as the most difficult issue in our domain
2. Some story components may be easily reused
3. If they cannot be reused but seems applicable, the system can ask the modeler to provide circumstantial distinction
constructivist acquisition model
The real issue? the HCI interface!
Discussion
• Petri Nets bring local causation and concurrency. Perfect match with our story-model.
• In addition, they provide good visualization, and model execution for debugging purposes (story animation).