... 2006-007 and Stanford KSL Technical Report KSL-06-03

1
Explanation Infrastructure Supporting
Transparency and Accountability

• Deborah L. McGuinness
• Co-Director and Senior Research Scientist
• Knowledge Systems, AI Laboratory, Stanford
  University
• Joint work with Li Ding, Cynthia Chang, Vasco
  Furtado, Paulo Pinheiro da Silva
• Inference Web is joint work with Richard Fikes,
  Alyssa Glass, Honglei Zeng, Mayukh Bhaowal, Bill
  Millar, Dhyanesh Narayan, Priyendra Deshwal,
• TAMI/Portia Privacy and Accountability Workshop
• 28-29 June 2006, Cambridge, MA USA

2
General Motivation
Provide interoperable knowledge provenance
infrastructure that supports explanations of
sources, assumptions, learned information, and
answers as an enabler for trust.

• Interoperability as more systems are use varied
  sources and multiple information manipulation
  engines, they benefit more from encodings that
  are shareable and interoperable
• Provenance - if users (humans and agents) are to
  use and integrate data from unknown, unreliable,
  or multiple sources, they need provenance
  metadata for evaluation
• Explanation/Justification if information has
  been manipulated (i.e., by sound deduction or by
  heuristic processes), information manipulation
  trace information should be available
• Trust if some sources are more trustworthy than
  others, representations should be available to
  encode, propagate, combine, and (appropriately)
  display trust values

3
TAMI Background

• Overall TAMI Ambition Explore privacy
  implications of semantic web technologies and
  determine viability.
• Goal Assess applicability of a usage limitation
  model (as opposed to or in addition to a data
  collection model) to data mining/profiling
  applications.
• Explore technical challenges of provable
  accountability with explicit justifications in
  large scale, heterogeneous information systems
  (i.e., the Web).
• Develop public policy models to encourage
  transparent, accountable data mining

4
Privacy and Explanation

• As more online data becomes available and as
  inference and interoperability increases, privacy
  protection will have to rely more on usage
  limitation rules and less on collection
  limitation rules
• Usage Limits depend upon
• Transparency knowledge provenance history of
  sources, data manipulations, and inferences is
  maintained in an interoperable form. Explanation
  technology used to allow examination of knowledge
  provenance by authorized parties (who may be the
  general public).
• Accountability ability to check whether the
  policies that govern data manipulations and
  inferences were in fact adhered to. Explanation
  technology used to examine inference usage.

5
Inference Web Infrastructure
WWW
Toolkit
Trust computation
IWTrust
OWL-S/BPEL
SDS (DAML/SNRC)
Proof Markup Language (PML)
End-user friendly visualization
IW Explainer/ Abstractor
N3
CWM (TAMI)
Expert friendly Visualization
Trust
KIF
JTP (DAML/NIMD)
IWBrowser
search engine based publishing
Justification
SPARK-L
SPARK (CALO)
IWSearch
Provenance
provenance registration
Text Analytics
IWBase
UIMA (NIMD/Exp Agg)
Inference Web Framework for explaining question
answering tasks by abstracting, storing,
exchanging, combining, annotating, filtering,
segmenting, comparing, and rendering proofs and
proof fragments provided by question answerers.
6
PML Ontology
7
How PML Works
isQueryFor
IWBase
Question fooquestion1 (how is arrest1
classified?)
Query fooquery1 (type arrest1 ?x)
hasAnswer
hasLanguage
Justification Trace
NodeSet foons1 (hasConclusion )
Language
hasInferencEngine
fromQuery
isConsequentOf
InferenceEngine
hasRule
InferenceStep
InferenceRule
hasAntecendent
Source
NodeSet foons2 (hasConclusion )

hasVariableMapping
Mapping
isConsequentOf
fromAnswer
hasSourceUsage
hasSource
SourceUsage
InferenceStep
usageTime
8
TAMI Architecture
Assertions with provenance
Action
Inference Engine(s) Truth Maintenance System
Proof Generation Services (Inference Web)
Justification Generator (why action does/doesnt
comply with policy)
9
Privacy Act background
10
Privacy Act
11
Sharing Restrictions

• (b) Conditions of Disclosure.
• No agency shall disclose any record which is
  contained in a system of records
• by any means of communication to any person or
  agency
• except . . . with the prior written consent of,
  the individual to whom the record pertains,
• unless disclosure of the record would be
• .
• (3) for a routine use 5 USC 552a(b)(3)

12
Routine Use

• as defined in subsection (a)(7)
• the use of such record for a purpose which is
  compatible with the purpose for which it was
  collected 5 USC 552a(a)(7)
• described under subsection (e)(4)(D)
• publish in the Federal Register upon
  establishment or revision a notice of the
  existence and character of the system of records,
  which notice shall include 5 USC 552a(e)(4)
• each routine use of the records contained in the
  system, including the categories of users and the
  purpose of such use 5 USC 552a(e)(4)(d)

13
One Simple Description
General Categories
Structured Components
14
Sample Code
Can share with FBI

• RU1 a RoutineUse
• recipient FBI
• category DataCategory3
• purpose CounterTerrorism
• dcdescription "May share where TSA becomes
  aware of information that may be related to an
  individual identified in the TSDB.
• DataCategory3 a DataCategory
• dcdescription "Information about people who
  are known or reasonably suspected to be or have
  been engaged in conduct constituting, in
  preparation for, in aid of, or related to
  terrorism.".

Data about possible terrorists
So they can investigate whether a criminal law
has been violated
15
Explanation

• Why was data sharing xyz acceptable?
• Using RoutineUse459
• We shared with the FBI
• Data belonging to DataCategory3
• For the purpose of CounterterrorismCriminalLawEnfo
  rcement
• Provenance RoutineUse459 is derived from
• SORN (statement of records notice) for Secure
  Flight
• Published at 70 FR 36319
• Encoded by DHS Office of the Privacy Officer

16
Explanation

• Why was data sharing xyz unacceptable?
• We checked all of the RoutineUses
• We shared with the IRS
• Data belonging to DataCategory3
• IRS is not authorized to receive DataCategory3
• For the purpose of CounterterrorismCriminalLawEnfo
  rcement
• Provenance all RoutineUses are derived from
• SORN for Secure Flight
• Published at 70 FR 3620
• Encoded by DHS Office of the Privacy Officer

17
Scenario 3 Examples(nerd level)

18
IWBrowser - Browse Debug
19
IWBrowser - Provenance
20
IWBrowser - Explanation
21
A fragment of PML data

• ltiwNodeSet rdfabout"ns__g463"gt
• ltiwhasConclusiongt _at_prefix
  60http//dig.csail.mit.edu/TAMI/lkagal/scenario
  3/rules62 .
• _at_prefix data 60http//dig.csail.mit.edu/TA
  MI/cph/v2/data.ttl62 .
• dataarrest-1 a NotJustifiedArrest
  charge 60http//dig.csail.mit.edu/TAMI/law/USC
  -18-22862 .
• lt/iwhasConclusiongt
• ltiwhasEnglishStringgtdataarrest-1 is-a
  NotJustifiedArrestand has charge
  60http//dig.csail.mit.edu/TAMI/law/USC-18-228
  62 .lt/iwhasEnglishStringgt
• ltiwhasLanguage rdfresource"http//infer
  enceweb.stanford.edu/registry/LG/N3.owlN3"/gt
• ltiwisConsequentOf rdfresource"is__g463
  "/gt
• lt/iwNodeSetgt
• ltiwInferenceStep rdfabout"is__g463"gt
• ltiwhasAntecedent rdfresource"ns__g419"
  /gt
• ltiwhasAntecedent rdfresource"ns__g420"
  /gt
• ltiwhasAntecedent rdfresource"ns__g425"
  /gt
• ltiwhasAntecedent rdfresource"ns__g479"
  /gt
• ltiwhasAntecedent rdfresource"ns__g502"
  /gt
• ltiwhasInferenceEngine rdfresource"http
  //inferenceweb.stanford.edu/registry/IE/CWM.owlCW
  M"/gt
• ltiwhasRule rdfresource"http//inference
  web.stanford.edu/registry/DPR/GMP.owlGMP"/gt
• ltiwhasVariableMapping rdfparseType"Reso
  urce"gt
• ltrdftype rdfresource"http//inferen
  ceweb.stanford.edu/2004/07/iw.owlMapping"/gt

22
Next Generation Browser
23
Discussion

• Exploring an explainable usage limitation model
  supported by semantic technologies
• Status
• Use case written up and (very recently) encoded
• Early prototype in place JUST integrated with
  CWM with use case 3 refinement in process
• Exploiting domain independent explanation tools
• Exploring options that leverage knowledge of N3
  and CWM supporting more useful tools/APIs/interfac
  es
• Exploring options that leverage knowledge of law
  supporting context-sensitive and appropriate
  presentations

24
Summary

• Leverage points include
• PML justification, provenance, and trust
  interlingua
• IW tools for interactive browsing, summarizing,
  searching, validation, abstraction, trust
  representation, propagation, presentation
  (domain independent)
• Experience explaining a wide variety of reasoners
  (JTP, SNARK, ), task processing engines (SPARK),
  learners (TAILOR), text analytics (UIMA), web
  services (SDS/BPEL),
• Continuing work
• Identify explanation (representation and
  presentation) requirements
• Refining registration and presentation
• Designing special purpose filters and
  abstractions
• Impact
• Interoperable justifications supporting
  transparency and accountability
• Has the potential to change publishing law (with
  markup) as well as presenting judgments (with
  interactive justifications)
• Audit trace

25
More Information

• Links
• http//iw.stanford.edu/ Inference Web home
• http//iw.stanford.edu/doc/project/tami/
  TAMI_at_Stanford
• http//dig.csail.mit.edu/TAMI TAMI home
• Papers
• (Inference Web) Deborah L. McGuinness and Paulo
  Pinheiro da Silva. Explaining Answers from the
  Semantic Web The Inference Web Approach. Journal
  of Web Semantics. Vol.1 No.4., pages 397-413,
  2004
• (PML) Paulo Pinheiro da Silva, Deborah L.
  McGuinness and Richard Fikes. A Proof Markup
  Language for Semantic Web Services. Information
  Systems. Volume 31, Issues 4-5, pages 381-395,
  2006
• (TAMI) Daniel J. Weitzner, Hal Abelson, Tim
  Berners-Lee, Chris P. Hanson, Jim Hendler, Lalana
  Kagal, Deborah L. McGuinness, Gerald J. Sussman,
  K. Krasnow Waterman. Transparent Accountable
  Inferencing for Privacy Risk Management.
  Proceedings of AAAI Spring Symposium on The
  Semantic Web meets eGovernment. AAAI Press,
  Stanford University, USA 2006. Also available as
  MIT CSAIL Technical Report-2006-007 and Stanford
  KSL Technical Report KSL-06-03.

26
Extras

27
IWBrowser - Browse Debug
28
IWBrowser - Provenance
29
IWBrowser - Explanation
30
A fragment of PML data

• ltNodeSet rdfabout"ns__g208"gt
• lthasConclusiongt _at_prefix
  60http//dig.csail.mit.edu/TAMI/cdk/scenario3/d
  ata.n362 .
• _at_prefix tr 60http//dig.csail.mit.edu/TAMI
  /cdk/scenario3/rules.n362 .
• Arrest05NY2343CRJFC a trJustifiedArrest
  . lt/hasConclusiongt
• lthasEnglishStringgtArrest05NY2343CRJFC
  is-a trJustifiedArrest .lt/hasEnglishStringgt
• lthasLanguage rdfresource"http//inferenc
  eweb.stanford.edu/registry/LG/N3.owlN3"/gt
• ltisConsequentOf rdfresource"is__g208"/gt
  
• ltreastep rdfresource"tami-scenario3-pro
  of.n3_g_L23C26"/gt
• lt/NodeSetgt
• ltInferenceStep rdfabout"is__g208"gt
• lthasAntecedent rdfresource"ns__g186"/gt
• lthasAntecedent rdfresource"ns__g187"/gt
• lthasAntecedent rdfresource"ns__g188"/gt
• lthasAntecedent rdfresource"ns__g207"/gt
• lthasInferenceEngine rdfresource"http//infe
  renceweb.stanford.edu/registry/IE/CWM.owlCWM"/gt
• lthasRule rdfresource"http//inferenceweb
  .stanford.edu/registry/DPR/GMP.owlGMP"/gt
• lthasVariableMapping rdfparseType"Resourc
  e"gt
• ltrdftype rdfresource"http//inferen
  ceweb.stanford.edu/2004/07/iw.owlMapping"/gt
• ltmapFromgthttp//people.csail.mit.edu/l
  kagal/tami/tami-scenario3-filter.n3Alt/mapFromgt

31
Next Generation Browser
32
PML based Explanation Adding trust-tab to
Wikipedia

33
The Original Wikipedia Article
34
Trust Tab

• Multiple Trust Tab
• citation based
• revision history based

Explanation about a fragment (author, trust value)
Fragments colored according trust value
35
PML Tabhttp//inferenceweb.stanford.edu/2006/02/e
xample1-iw-wiki.owl
fragment
ltiwNodeSet rdfabout"http//foto.stanford.edu/me
diawiki-1.4.12/index.php/Natural_number"gt ltIn
mathematics, a natural number is either a
positive integer lt/iwhasConclusiongt
ltiwhasLanguage rdfresource"http//inferenceweb.
stanford.edu/registry/LG/English.owlEnglish"/gt
ltiwisConsequentOfgt ltiwInferenceStepgt
ltiwhasRule rdfresource"http//inferenceweb.st
anford.edu/registry/DPR/Told.owlTold"/gt
ltiwhasInferenceEngine rdfresource"http//infere
nceweb.stanford.edu/registry/IE/CitationTrust.owl
CitationTrust"/gt ltiwhasSourceUsagegt
ltiwSourceUsagegt ltiwhasSourcegt
ltiwSource rdfabout"http//inferenceweb
.stanford.edu/wp/registry/PER/Alexandrov.owlAlexa
ndrov"/gt lt/iwhasSourcegt
lt/iwSourceUsagegt lt/iwhasSourceUsagegt
lt/iwInferenceStepgt lt/iwisConsequentOfgt lt/iw
NodeSetgt ltiwAggregatedTrustRelationgt
ltiwhasTrustingParty rdfresource"http//inferenc
eweb.stanford.edu/wp/registry/ORG/wikipedia.owlwi
kipedia"/gt ltiwhasTrustedParty
rdfresource"http//foto.stanford.edu/mediawiki-1
.4.12/index.php/Natural_number"/gt
ltiwhasTrustValue rdfdatatype"http//www.w3.org/
2001/XMLSchemafloat"gt0.1766lt/iwhasTrustValuegt lt/
iwAggregatedTrustRelationgt ltiwAggregatedTrustRe
lationgt ltiwhasTrustingParty
rdfresource"http//inferenceweb.stanford.edu/wp/
registry/ORG/wikipedia.owlwikipedia"/gt
ltiwhasTrustedParty rdfresource"http//inference
web.stanford.edu/wp/registry/PER/Alexandrov.owlAl
exandrov"/gt ltiwhasTrustValue
rdfdatatype"http//www.w3.org/2001/XMLSchemaflo
at"gt0.1766lt/iwhasTrustValuegt lt/iwAggregatedTrust
Relationgt
fragment trust
author trust
36
PML based Explanation Explaining Cognitive
Assistants DARPA PAL Programs CALO system

37
Explanation Process

• Initial request and answer strategy
• ltusergt Why are you doing ltsubtaskgt?
• ltsystemgt I am trying to do lthigh-level-taskgt and
  ltsubtaskgt is one subgoal in the process.
• Follow-up questions for mixed initiative dialogue
• ltusergt Why are you doing lthigh-level-taskgt?
• ltusergt How did you learn to do lthigh-level-taskgt
  in this way?
• ltusergt Why havent you completed ltsubtaskgt yet?
• ltusergt Why is ltsubtaskgt a subgoal of
  lthigh-level-taskgt?
• ltusergt When will you finish ltsubtaskgt?
• ltusergt What sources did you use to do ltsubtaskgt?

McGuinness, D.L. Pinheiro da Silva, P. Glass,
A. Wolverton, M. Explaining Task Processing in
Cognitive Assistants. 2006. Technical Report,
KSL-06-06, Knowledge Systems Lab., Stanford.
38
CALO TM Explainer UI
Initial explanation, with links
indicating follow-up queries and alternate
strategies.