Web Explanations for Semantic Heterogeneity Discovery

1
Web Explanations for Semantic Heterogeneity
Discovery
Pavel Shvaiko
work in collaboration with Fausto
Giunchiglia, Paulo Pinheiro da Silva and Deborah
L. McGuinness
2nd European Semantic Web Conference (ESWC), 1
June 2005, Crete, Greece
2
Outline

• Introduction
• Semantic Matching
• Inference Web (IW) Framework
• Explaining Semantic Matching using IW
• Experimental Study
• Conclusions

3
Introduction

• Information sources (e.g., database schemas,
  classifications or ontologies) can be viewed as
  graph-like structures containing terms and their
  inter-relationships
• Matching is one of the key operations for
  enabling the Semantic Web since it takes two
  graph-like structures and produces a mapping
  between the nodes of the graphs that correspond
  semantically to each other
• Matching, however, requires explanations because
  mappings between terms are not always intuitively
  obvious to human users

4

• Semantic Matching

5
Semantic Matching
Semantic Matching Given two graphs G1 and G2,
for any node n1i ? G1, find the strongest
semantic relation R holding with node n2j ? G2
We compute semantic relations by analyzing the
meaning (concepts, not labels) which is codified
in the elements and the structures of
schemas/classifications
Technically, labels at nodes written in natural
language are translated into propositional
logical formulas which explicitly codify the
labels intended meaning. This allows us to
codify the matching problem into a propositional
validity problem
6
Example Two simple classifications
A1
A2
D.E.

• Axioms ? rel (Context1, Context2)

(Images1?Pictures2) ? (Europe1?Europe2) ?
(Images1 ? Europe1) ? (Europe2 ?Pictures2)
7
S-Match
Expl.
8

• Inference Web (IW) Framework

9
The IW Framework Overview
Inference Web is a framework enabling
applications to generate portable and distributed
explanations for their answers
10

• Explaining Semantic Matching
• using IW

11
Producing Explanations

• In order to explain mappings produced by S-Match
  and thereby increase the trust level of its
  users, we need to provide information about
• background theories (e.g., WordNet)
• JSAT manipulations of propositional formulas

WordNet
12
Default Explanation
A default explanation of mappings the S-Match
system produces is a short, natural language,
high-level explanation without any technical
details. It is designed to be intuitive and
understandable by ordinary users Query find
"European pictures"
Query
13
Explaining Background Knowledge
Suppose that the agent still does not trust the
answer and may want to see the sources of
metadata information behind the mapping
14
Explaining Logical Reasoning
If the mappings derivation process needs to be
explained, using the JSAT SAT engine, S-Match
produces a trace of the DPLL procedure
15
Experimental Study
16
Preliminary Results
Goal to obtain a vision of how the S-Match
explanations potentially scale to requirements of
the Semantic Web
17
Conclusions

• We use the Proof Mark-up Language for
  representing S-Match proofs, thus
  facilitating interoperability
• We use meaningful terms rather than numbers in
  the DIMACS format, thus facilitating
  understandability
• We use the IW tools, thus facilitating
  customizable, interactive proof and explanation
  presentation and abstraction
• Our solution is potentially scalable to the
  Semantic Web requirements

18
Future Work

• Developing an environment, which efficiently
  exploits the IW proofs and explanations, in order
  to make the S-Match matching process
  (fully-fledged) interactive and iterative
• Improving the S-Match proofs and explanations by
  using abstraction techniques more extensively
• Conducting a user satisfaction study of the
  explanations
• Extending explanations to other SAT engines as
  well as to other non-SAT DPLL-based inference
  engines

19
References

• Project website at DIT - ACCORD
  http//www.dit.unitn.it/accord/
• Project website at KSL - IW http//iw.stanford.ed
  u/
• F. Giunchiglia, P. Shvaiko Semantic matching.
  The Knowledge Engineering Review Journal,
  18(3)265-280, 2003.
• F. Giunchiglia, P. Shvaiko, M. Yatskevich
  S-Match an algorithm and an implementation of
  semantic matching. In Proceedings of ESWS, pages
  61-75, 2004.
• D. McGuinness, P. Pinheiro da Silva Explaining
  Answers from the Semantic Web The Inference Web
  Approach. Journal of Web Semantics, 1(4) 397-
  413, 2004.

20

• Thank you!