Get documents about "Explanation Paulo Pinheiro da Silva
Shared by: liaoqinmei
-
Stats
- views:
- 1
- posted:
- 11/5/2012
- language:
- Unknown
- pages:
- 48
Document Sample
Explainable Systems:
The Inference Web Approach
Paulo Pinheiro da Silva
Stanford University
In collaboration with Deborah L. McGuinness, Richard E. Fikes,
Cynthia Chang, Priyendra Deshwal, Dhyanesh Narayanan, Alyssa
Glass, Selene Makarios, Jessica Jenkins, Bill Millar, Eric Hsu and
many people from IBM, SRI, ISI, IHMC, U. Toronto, U. Trento, U.
Fortaleza, U. Texas Austin, Rutgers U., Maryland U., Batelle,
SAIC, UCSF, MIT W3C
Overview
1. What are explainable systems
and why should we care about
them?
2. Inference Web: Enabling Explainable Systems
3. Explainable Systems in Action
4. Explainable Systems 10 years from now
Paulo Pinheiro da Silva
Explanation Need
I need to send Google-2.0,
Paulo a letter where is
but I don’t Paulo’s office? Google-2.0, why is
know his Paulo’s address
address. “Manchester, UK”?
I believe Paulo
lives in the U.S.
So, Stanford,
CA, USA.
appears to be a
possible answer.
[Betty]
Paulo Pinheiro da Silva
Explanation in Action
Why should I Why should I OK, “Manchester, UK” was
believe this? believe these? Paulo’s address in May, 2002 and
we are in 2005 !!
I’ll send his letter to Stanford.
Paulo At Manchester, UK
[Betty]
transitivity of At
Paulo At University of Manchester University of Manchester At Manchester, UK
Source: http://www.cs.man.ac.uk/~pinheirp Source: http://www.cs.man.ac.uk
Source usage: May/2002 Source usage: May/2002
Paulo Pinheiro da Silva
What are Explainable Systems?
question
question
answer
answer explanation question answer
explanation
request 1
request 1
expl. 1 explanation explanation
request 1 request 1 expl. 1
explanation 1
…
…
…
expl. n answer
explanation
understanding explanation expl. n
request n
request n
[Bob] explanation n
Paulo Pinheiro da Silva
Why should we care about
explainable systems?
As system users, we often need:
To understand system’s response
To trust system’s responses
Many explanation concerns are the same
as in early systems such as
Shortliffe’s MYCIN [1976]
Swartout’s XPLAIN [1983]
Paulo Pinheiro da Silva
Why should we care about
explainable systems even more now?
Systems are far more complex than 30
years ago
Hybrid and distributed processing, e.g., web services, the
Grid
Large number of heterogeneous, distributed information
sources, e.g., the Web
More variation in reliability of information sources, e.g.,
information extraction
Sophisticated information integration methods, e.g.,
SIMS, TSIMMIS
Now we have less understanding (and
sometimes less trust) of system’s answers
and behavior
Now we have even more reasons for
systems to explain their responses
Paulo Pinheiro da Silva
How to Enable Explainable Systems?
1 -> ((allof (the played-by of (the instances
of Project-Leader)) where
(It isa Person)) = (:set *Helen *Jody)) Which
2 -> (allof (the played-by of (the instances
of Project-Leader)) where
information do
(It isa Person))
question answer I have to
3 -> (forall (the played-by of (the
instances of Project-Leader)) generate an
where (It isa Person) It)
4 -> (the played-by of (the instances of explanation explanation?
Project-Leader))
5 -> (the instances of Project-Leader) request 1 expl. 1
5 (1) Local value(s): (:set *COGS-Proj-
Leader-1
…
*HI-LITE-ProjectLeader-1 *SKIPR-
ProjectLeader-1)
6 -> (:set *COGS-Proj-Leader-1 *HI-LITE- explanation I may have (or
ProjectLeader-1
*SKIPR-ProjectLeader-1) [for (the request n expl. n may be able to
instances of Project-Leader)]
6 <- (*COGS-Proj-Leader-1 *HI-LITE-
record) data
ProjectLeader-1 describing how I
*SKIPR-ProjectLeader-1) [(:set...
5 (2) From inheritance: (:set *COGS- manipulate
Proj-Leader-1
*HI-LITE-ProjectLeader-1 *SKIPR- information to
ProjectLeader-1)
produce answers!
Paulo Pinheiro da Silva
Explainable System Challenge
Explanation
Understanding Trust
The
GAP
Information Manipulation Data
Paulo Pinheiro da Silva
Overview
1. What are explainable systems and why
should we care about them?
2. Inference Web: Enabling
Explainable Systems
3. Explainable Systems in Action
4. Explainable Systems 10 years from now
Paulo Pinheiro da Silva
Requirements for Explainable Systems
Information Manipulation Traces
hybrid, distributed, portable, shareable, combinable encoding of
proof fragments supporting multiple justifications
Presentation
multiple display formats supporting browsing, visualization, etc.
Abstraction
understandable summaries
Interaction
multi-modal mixed initiative options including natural-language and
GUI dialogues, adaptive, context-sensitive interaction
Trust
source and reasoning provenance, automated trust inference
[McGuinness & Pinheiro da Silva, ISWC 2003,
J. Web Semantics 2004]
Paulo Pinheiro da Silva
Explainable System Challenge
Explanation
Proof Markup Language
Information
Manipulation
Data
Paulo Pinheiro da Silva
Proof Markup Language:
Node Sets and Inference Steps
Direct Direct Direct
Assertion Assertion Assertion
From KB1 From Doc1 from Doc2
A->(A^B) A B A DAG of
PML Node Sets
Modus Direct
(a collection
AND Assertion (DA) of justifications)
Ponens
Intro (^I) from KB1
(MP)
A^B
Extracted
A->(A^B) A A B DA Proofs for the
MP ^I
A^B conclusion A^B
A^B A^B
Paulo Pinheiro da Silva
Encoding Hybrid and Distributed
Proof Fragments
Proof Markup Language has a web-based solution for
distribution
Specification written in W3C’s OWL
Each node set has one URI
Node sets can be used to combine proofs generated by
multiple agents
OMEGA [Siekmann et al.,CADE2002] has a nice solution
for hybrid proofs
http://foo.com/NS.owl#NS124 http://bar.com/NS.owl#NS125
rule: Modus Ponens (MP)
hasEngine: JTP
conclusion: (and A B)
A^B
hasLanguage: KIF
http://foo.com/NS.owl#NS123
Paulo Pinheiro da Silva
Information Manipulation Traces
Proof Markup Language
Information
Differences Formal Proofs
manipulation traces
Optional use or use of
Use of rules Mandatory
‘unregistered rule’
Written in some formal Written in a formal or
Sentences language (e.g., KIF, CL, informal language including
DIMACS, etc.) natural language
Use of multiple
representation Uncommon Common
languages
Proof Markup Language covers the full
spectrum of information manipulation traces!
[Pinheiro da Silva, McGuinness & Fikes, IS 2005]
Paulo Pinheiro da Silva
Explainable System Challenge
Explanation
Proof Markup Language
Information
Provenance
Manipulation
Meta-data
Data
Paulo Pinheiro da Silva
Infrastructure: IWBase
Meta-data useful for disclosing knowledge
provenance and reasoning information such as
descriptions of
inference engines along with their supported inference
rules
Information sources such as organizations, publications
and ontologies
Languages along with their axioms
Core IWBase as well as domain IWBases
OWL files for interoperability and database for
scaling
[McGuinness & Pinheiro da Silva, IIWeb 2003]
Paulo Pinheiro da Silva
Infrastructure: Core IWBase
Statistics for relevant
domain independent
meta-data:
Inference Engines 29
Axioms 56
Declarative Rules 38
Method Rules 10
Derived Rules 6
Languages 12
Paulo Pinheiro da Silva
Explainable System Challenge
Explanation
Presentation
Proof Markup Language
Information
Provenance
Manipulation
Meta-data
Data
Paulo Pinheiro da Silva
Browsing Proofs (1/2)
Enable the visualization of proofs (and abstracted proofs)
Proofs can be “extracted” and browsed from both local and
remote PML node sets and can be combined
Links provide access to proof-related meta-information
Paulo Pinheiro da Silva
Browsing Proofs (2/2)
Paulo Pinheiro da Silva
Explainable System Challenge
Explanation
Presentation
Abstraction
Proof Markup Language
Information
Provenance
Manipulation
Meta-data
Data
Paulo Pinheiro da Silva
Knowledge Provenance Elicitation
Google-2.0 says
Provenance information may be ‘A^B’ is the answer
essential for users to trust answers. for my question.
Data provenance (aka data lineage) is “has opinion” “has opinion”
Why should I
defined and studied in the database “has opinion” believe this?
literature. BBC NYT CNN
[Buneman et al., ICDT 2001]
[Cui and Widom, VLDB 2001]
DA DA DA
Knowledge provenance extends A->(A^B) A B
data provenance by adding data
derivation provenance information
MP ^I DA
[Pinheiro da Silva, McGuinness & A^B
McCool, Data Eng. Bulletin, 2003]
A->(A^B) A A B
MP ^I Dir.Ass.
A^B A^B A^B
(CNN,BBC) (BBC,NYT) (CNN)
Paulo Pinheiro da Silva
Knowledge Provenance Example
Paulo Pinheiro da Silva
Abstracting Proofs
Explanation tactics (a.k.a. rewriting rules) may be
used to abstract proofs into more understandable
and manageable explanations
Enable the use of axioms as inference rules
preventing the presentation of primitive (and
potentially less interesting and useful) rules
Eliminate intermediate results from proofs
Paulo Pinheiro da Silva
Abstracting Proofs: An Example (1/2)
Direct assertion
(implies Direct assertion Direct assertion
Direct assertion
(and (Holds (owner ?person (Holds (owner (organization Direct assertion
(Holds (owner
?object) ?when) JoesephGradgrind GradgrindFoods)
JoesephGradgrind (organization
(organization ?object)) GradgrindFoods) Apr1_03) GradgrindFoods) Apr1_03) GradgrindFoods)
(Holds* (hasOffice ?person
?object) ?when)) Assumption
Generalized Modus Ponens
Direct assertion (not (Ab (hasOffice Organization Owner Typically Has
(Holds* (hasOffice JosephGradgrind Office at Organization
(implies
JoesephGradgrind ?where) ?when))
(and (Holds* ?f ?t)) (Holds (hasOffice
GradgrindFoods) Apr1_03)
(not (Ab ?f ?t)) JoesephGradgrind
(Holds ?f ?t)) GradgrindFoods) Apr1_03)
Generalized Modus Ponens
(Holds ((hasOffice
Tactic ABSTRACTED PROOF
JoesephGradgrind Library
GradgrindFoods) Apr1_03)
Explanation tactic: “Organization Owner
Typically Has Office at Organization”
Abstractor algorithm
(implies
1) Match conclusion (key for
Direct assertion
(and (Holds (owner ?person
?object) ?when)) (Holds ((owner ?person Direct assertion selecting tactics)
(organization ?object)) ?object) ?when)
(organization ?object) 2) Match leaf nodes
(Holds* (hasOffice ?person
?object) ?when))
3) Unify
(implies
Generalized Modus Ponens 4) Propagate conclusion
(Holds* ((hasOffice ?person
(and (Holds* ?f ?t)) (not (Ab (hasOffice
(not (Ab ?f ?t))
?object) ?when)
?person ?object) 5) Apply the assertion-level rule
(Holds ?f ?t)) ?when))
6) Propagate justified nodes
Generalized Modus Ponens
(Holds ((hasOffice ?person
?object) ?when) Paulo Pinheiro da Silva
Abstracting Proofs: An Example (2/2)
Direct assertion A rule says that
(Holds (owner the owner of an organization
JoesephGradgrind Direct assertion typically has an office in an
GradgrindFoods) (organization organization
Apr1_03) GradgrindFoods) Because
• JosephGrardgrind owned
GradgrindFoods on April 1st 2003
Organization Owner Typically • GradgrindFood is an organization
Has Office at Organization therefore
(Holds (hasOffice • JosephGradgrind had an office at
JoesephGradgrind GradgrindFoods on April 1st, 2003.
GradgrindFoods) Apr1_03)
ABSTRACTED PROOF IN
DISCURSIVE STYLE
ABSTRACTED PROOF
Assertion-level rules are introduced Maybury describes strategies for
in [Huang, PRICAI 1996]. rewriting abstracted proofs into
English [AAAI 1991, AAAI 1993].
Explanation tactics supports
multi-level abstraction of proofs
Paulo Pinheiro da Silva
Explainable System Challenge
Explanation
Understanding
Interaction
Presentation
Abstraction
Proof Markup Language
Information
Provenance
Manipulation
Meta-data
Data
Paulo Pinheiro da Silva
Explaining Answers: GUI Explainer
Users can exit the explainer
providing feedback about
their satisfiability with
explanation(s)
Users can ask for
alternative explanations
Paulo Pinheiro da Silva
Explainable System Challenge
Explanation
Understanding
Interaction
Presentation
Abstraction
Inference
Proof Markup Language
Meta-Language
Information Inference
Provenance
Manipulation Rule
Meta-data
Data Specs
Paulo Pinheiro da Silva
Inference Meta Language (InferenceML)
An inference rule involves pattern of transformations
on expressions to produce a conclusion
InferenceML uses schemas to state such
transformations
InferenceML defines a schema to be a pattern, which
is any expression of CL in which:
some lexical items have been replaced by a schematic
variable (or meta-variable)
Example:
ndUI: '(forall (' N ')' q ')' |- ' (forall (' N - N.i ')' q[t/N.i] ')';; (Name N) (Sent q) (Term t)
Paulo Pinheiro da Silva
Checking Proofs
DA DA
(A) (implies (A)
(and A B))
From
IWBase
MP
(and A B)
MP: x; '(implies ' x y ')' |- y ;; (Sent x y)
(A) ; (implies (A) (and A B)) |- (and A B)
binding of expressions to schematic variables:
• x binds to (A)
• y binds to (and A B)
the rule schema instantiates directly to:
=
(A) ; (implies (A) (and A B)) |- (and A B)
Paulo Pinheiro da Silva
Explainable System Challenge
Explanation
Understanding Trust
Interaction
Presentation
Abstraction
Inference
Proof Markup Language Meta-Language
Information Inference
Provenance
Manipulation Rule
Meta-data
Data Specs
Paulo Pinheiro da Silva
IWTrust: Trust in Action
Google-2.0 says
‘A^B’ is the answer
Trust can be
for my question.
inferred from a ++
Web of Trust. Why should
? ++ 0 I trust the
IWTrust provides infrastructure + answer?
for building webs of trust. XYZ NYT CNN
The infrastructure includes a
DA DA DA
trust component responsible for
computing trust values for A->(A^B) A B
answers.
IWTrust is described in MP ^I DA
[Zaihrayeu, Pinheiro da A^B
Silva & McGuinness,
iTrust 2005]
A->(A^B) A A B
MP ^I DA
A^B A^B B
A^B
0 (CNN,XYZ) +
? ? (XYZ,NYT) ++
+ (CNN) 0
Paulo Pinheiro da Silva
Inference Web and Paulo
Paulo is a co-technical leader of the Inference Web
project
Paulo was the main IW developer during 1 ½ years
Paulo has been the manager of the IW development
team including members with the following profile:
1 research programmer
3 masters students
1 Ph.D. student
Paulo has organized the IW weekly meetings
Paulo has been responsible for presenting and
demonstrating IW solutions at several DARPA and
ARDA PI meetings
Paulo has participated of the writing of grant
proposals
Paulo Pinheiro da Silva
Overview
1. What are explainable systems and why
should we care about them?
2. Inference Web: Enabling Explainable
Systems
3. Explainable Systems in Action
4. Explainable Systems 10 years from now
Paulo Pinheiro da Silva
Application Areas
Information extraction – IBM (UIMA), Stanford (TAP)
Information integration – USC ISI (Prometheus/Mediator); Rutgers
University (Prolog/Datalog)
Task processing – SRI International (SPARK)
Theorem proving
First-Order Theorem Provers –SRI International (SNARK); Stanford (JTP);
University of Texas, Austin (KM)
SATisfiability Solvers – University of Trento (J-SAT)
Expert Systems – University of Fortaleza (JEOPS)
Service composition – Stanford, University of Toronto, UCSF (SDS)
Semantic matching – University of Trento (S-Match)
Debugging ontologies – University of Maryland, College Park
(SWOOP/Pellet)
Problem solving – University of Fortaleza (ExpertCop)
Trust Networks – U. of Trento (IWTrust)
No single explanation approach has been used in so
many diversified areas as Inference Web!
Paulo Pinheiro da Silva
Extraction as Inference
Goal: To provide browsable justifications
of information extraction
Strategy: Reuse, adapt, and integrate
existing technology:
justification technology - Inference Web
extraction technology - IBM’s UIMA
Requires that systems to describe their
processing as logical inferences
Requires a new perspective: IE as Inference
[Murdock, Pinheiro da Silva et al., AAAI’s SSS 2005]
Paulo Pinheiro da Silva
Extraction As Inference:
An Example (1/2)
Solution: Direct assertion fromgradgrind.txt
Joseph Gradgrind is the owner
A taxonomy of extraction tasks expressed as inference of Gradgrind Foods
rules Entity Recognition
IBM EAnnotator
Components that record IE justifications using rules in Joseph Gradgrind is the owner
the taxonomy of Gradgrind Foods[organization]
Entity Identification
We have identified 9 types of extraction inferences: IBM Cross-Annotator Coreference
Joseph Gradgrind is the owner
6 for analysis, and 3 for integration of Gradgrind Foods[organization]
[refers to GradgrindFoods]
Direct assertion from KB1
Direct assertion from
(implies Direct assertion from KB1 KB1
Extracted Entity Classification
(and (Holds (owner ?person (Holds (owner Document
(organization Coreference
?object) ?when) JoesephGradgrind GradgrindFoods)
(organization GradgrindFoods)
(organization ?object)) GradgrindFoods) Apr1_03)
(Holds* (hasOffice ?person
?object) ?when)) Assumption
Generalized Modus Ponens
Direct assertion from KB1 (not (Ab (hasOffice
(Holds* (hasOffice JosephGradgrind
(implies
JoesephGradgrind ?where) ?when))
(and (Holds* ?f ?t))
GradgrindFoods) Apr1_03)
(not (Ab ?f ?t))
(Holds ?f ?t))
Generalized Modus Ponens
(Holds ((hasOffice
JoesephGradgrind
GradgrindFoods) Apr1_03)
Paulo Pinheiro da Silva
Extraction As Inference:
An Example (2/2)
Why should I Why should I Why should I believe that
believe this? believe these? these documents say that?
Paulo At Manchester, UK
[Betty]
transitivity of At
Theorem
Proving
Paulo At University of Manchester University of Manchester At Manchester, UK
http://www.cs.man.ac.uk/~pinheirp http://www.cs.man.ac.uk Information
Extraction
Paulo is a PhD student at University of Manchester. University of Manchester is located in Manchester, UK.
Paulo Pinheiro da Silva
Explaining Tool Responses
Inferences for explaining
Requests and Responses
answers (aka beliefs), and
tasks (including actions)
Generalization
Inferences for explaining
Questions and Answers answers (aka beliefs)
Explain (v. tr.)1:
“To offer reasons for the actions, beliefs,
or remarks of (oneself).”
New perspective: Task processing as inference
1Dictionary.com
Paulo Pinheiro da Silva
NL Explainer: An Example
<user>: What are you doing now?
<system>: I am trying to get an approval to buy a
laptop.
<user>: Why?
[note: “Why?” is rephrased to “Why are you trying
to get an approval to buy a laptop?]
<system>: I have completed the previous
requirement to get quotes so I am now working on
get approval.
<user>: OK, I am happy with your explanation.
Levering explanation dialogues as in [Fiedler, IJCAI 2001]
Using natural language support as in [Allen et al., AAMAS 2002]
Paulo Pinheiro da Silva
Overview
1. What are explainable systems and why should
we care about them?
2. Inference Web: Enabling Explainable Systems
3. Explainable Systems in Action
4. Explainable Systems 10 years from
now
Paulo Pinheiro da Silva
Inference Web Contributions
1. Language for encoding
hybrid, distributed proof 6
fragments based on web
technologies. Support for Explanation
both formal and informal
proofs (information Understanding 5 Trust
manipulation traces). 4 Interaction
2. Support (registry,
4
language, services) for Presentation
knowledge provenance. 4 Abstraction
3. Declarative inference rule Inference
3
representation for checking Proof Markup Language Meta-Language
1 2
hybrid, distributed proofs. Information 2 Inference
Provenance
4. Multiple strategies for proof Manipulation Rule
Meta-data
Data Specs
abstraction, presentation
and interaction.
5. End-to-end trust value computation for answers.
6. Comprehensive solution for explainable systems.
Paulo Pinheiro da Silva
Open Issues
Automated generation of explanation tactics
Performance for abstracting and checking proofs
Use of machine learning and user modeling to
support interaction
Adaptive explanations
Explanation contexts
Modeling user knowledge
Metrics and evaluations for explainable systems
Paulo Pinheiro da Silva
Three Years From Now
An initial research community working on explainable
systems
Adaptive explanations based on user modeling
IWBase registration of a large set of software systems
Registration of a comprehensive set of primitive rules
Established library of explanation tactics
First generation of metrics and
evaluation methods for explainable
systems
Inference Web is a
solution for the Semantic
Web proof and trust layers
http://www.w3.org/2004/Talks/0412-RDF-functions/slide4-0.html
Paulo Pinheiro da Silva
Ten Years From Now
An established research community working on
explainable systems
A theory for explainable systems
Established metrics for explainable systems
First (or second) generation of industrial
explainable systems
A standard language for encoding information
manipulation traces (probably derived from PML
among other proposals). The language will include
support for the following:
probabilistic reasoning
inductive reasoning
Paulo Pinheiro da Silva
and Inference Web
Immediate connections
Explaining Task Processing
TaskTracer
CALO
with Intelligent Information Systems team
Explaining Tool Responses
Explaining WYSIWYT –
with End Users Shaping Effective Software team
Potential connections
Explanation generation
Filtering
Learning
Explanation-based learning
with Learning and Adaptive Systems team
Explaining pattern and object recognition from videos
and graphs
with Computer Graphics and Vision
Paulo Pinheiro da Silva
