66
Finding Semantic Matches Between Conceptual Graphs Peter Yeh May 14, 2002

Finding Semantic Matches Between Conceptual Graphs

  • Upload
    draco

  • View
    33

  • Download
    0

Embed Size (px)

DESCRIPTION

Finding Semantic Matches Between Conceptual Graphs. Peter Yeh May 14, 2002. Talk Outline. Motivation. Matching. Rewrite Rules. Matching in a KB. Elaboration. Applications. Future Work. Related Work. Motivation. - PowerPoint PPT Presentation

Citation preview

Page 1: Finding Semantic Matches Between Conceptual Graphs

Finding Semantic Matches Between Conceptual Graphs

Peter Yeh

May 14, 2002

Page 2: Finding Semantic Matches Between Conceptual Graphs

Talk Outline

• Motivation.

• Matching.

• Rewrite Rules.

• Matching in a KB.

• Elaboration.

• Applications.

• Future Work.

• Related Work.

Page 3: Finding Semantic Matches Between Conceptual Graphs

Motivation• Goal: Develop a matcher which can determine if two

concepts are semantically alike.• Problem: Discrepancies in representation.

"John's hand is in a jar filled with cookies."

Page 4: Finding Semantic Matches Between Conceptual Graphs

Motivation

• Why: A good semantic matcher has many useful applications – Rule Base: A rule firing requires a match of the

consequent or antecedent. – Knowledge Acquisition: Locating relevant pieces

of prior knowledge to accelerate knowledge entry. – Knowledge-Based IR: Retrieve information based

on semantics. – Pattern Completion: Locate relevant pieces of

knowledge to elaborate a user's concept.

Page 5: Finding Semantic Matches Between Conceptual Graphs

Pattern Completion

User Input

KB

Page 6: Finding Semantic Matches Between Conceptual Graphs

Pattern Completion

User Input

A piece of prior knowledge from

the KB.

KB

Page 7: Finding Semantic Matches Between Conceptual Graphs

Pattern Completion

The result from elaborating the user’s input

Page 8: Finding Semantic Matches Between Conceptual Graphs

Talk Outline

• Motivation.

• Matching.

• Rewrite Rules.

• Matching in a KB.

• Elaboration.

• Applications.

• Future Work.

• Related Work.

Page 9: Finding Semantic Matches Between Conceptual Graphs

Matching

• Problem: Given two concepts, are they semantically similar?

• Formally,

Given:C1: A concept. C2: A concept.c: A match criterion.

C1 and C2 semantically match iff C1 C2 and c is satisfied.

Page 10: Finding Semantic Matches Between Conceptual Graphs

Matching (cont.)

• A part of C1 and C2 intersect iff xx', yy', and rr'.

• The general problem is called subgraph morphism in the literature and is NP complete.

• We are matching labeled type graphs which is polynomial. However, the matching problem is embedded within other problems.

I

.

C1 C2

Page 11: Finding Semantic Matches Between Conceptual Graphs

Match Criterion

• C1 and C2 intersecting is not enough. The match criterion must also be satisfied.

• Match criterion defines what type of match is being performed.

• Criterions: – Exact match: C1 is either isomorphic to or a subgraph of

C2.

– Auto-Classification: The necessary conditions of C1 is a subgraph of C2 and the root of C1 subsumes the root of C2.

– Similarity match: The intersection of C1 and C2 is not empty.

Page 12: Finding Semantic Matches Between Conceptual Graphs

Talk Outline

• Motivation.

• Matching.

• Rewrite Rules.

• Matching in a KB.

• Elaboration.

• Applications.

• Future Work.

• Related Work.

Page 13: Finding Semantic Matches Between Conceptual Graphs

Rewrite Rules• We need rewrite rules to handle discrepancies

between two representations of the same piece of information.

• Rewrite rules are of the form LHS RHS.• The LHS and RHS are closely coupled. As a result, a

rewrite affects only that part of a concept which is an instantiation of the LHS.

• We envision two types of rewrites: – Sound rewrite rules. – Heuristic rewrite rules.

Page 14: Finding Semantic Matches Between Conceptual Graphs

Sound Rewrite Rules• Sound rewrites are universally true.• They are semantics preserving.• They exploit the meta-properties of relations:

– transitivity, symmetry, and reflexivity.

– part ascension and covers rule.• Our current set of rewrites is not exhaustive.• The methodology we use to populate our

library of rewrites is– Identify a pattern.

– Exhaustively fill out the pattern with all valid instantiations.

– Generalize when possible.

Page 15: Finding Semantic Matches Between Conceptual Graphs

Sound Rewrites: Transitivity

• Transitivity.

• 21 of our 97 relations are transitive.

Page 16: Finding Semantic Matches Between Conceptual Graphs

Sound Rewrites: Symmetry

• Symmetry.

• 6 of our 97 relations are symmetric.

Page 17: Finding Semantic Matches Between Conceptual Graphs

Sound Rewrites: Part Ascension

• Part Ascension.

• The set S of part-onomic relations is:– is-part-of

– subevent-of

– is-region-of

Page 18: Finding Semantic Matches Between Conceptual Graphs

Sound Rewrites: Covers

• Transitivity and part ascension fit a more general pattern that we call the covers rule.

Page 19: Finding Semantic Matches Between Conceptual Graphs

Sound Rewrites: Some More Covers Rule

relation Trans. Sym. Reflex. covers

causes(caused-by) X - -

subevent, resulting-state(subevent-of, resulting-from)

defeats(defeated-by) - - - (caused-by, subevent-of)

enables(enabled-by) X - -

causes, resulting-state, subevent(caused-by, resulting-from,

subevent-of)entails

(entailed-by) X - -causes, resulting-state, subevent

(caused-by, resulting-from,subevent-of)

inhibits(inhibited-by) - - -

resulting-state, subevent-of(caused-by, resulting-from,

subevent-of)by-means-of

(means-by-which) X - - -

prevents(prevented-by) - - -

subevent-of(subevent-of, caused-by,

resulting-from)

An excerpt of some of the covers rule from our rewrite library.

Page 20: Finding Semantic Matches Between Conceptual Graphs

Sound Rewrites: Some Statistics on Covers

• We have 97 relations in our slot language*

• Total number of valid xyz combinations where the range of r and the domain of r’ are the same is 2137.

• Total number of valid xyz combinations where y is within the range z is 791.

• Total number of covers rule is 210.• Percentages

– range of r and domain of r’ the same: 9.8%– y within the range of z: 26.5%

r r’

r r’

Page 21: Finding Semantic Matches Between Conceptual Graphs

Sound Rewrites: Complex Rules

• Sound rewrites can also capture complex relationships.

• For example:

Page 22: Finding Semantic Matches Between Conceptual Graphs

Sound Rewrites: Complex Rules

• The representation of the previous example

• This is an instantiation of the rewrite rule:

Page 23: Finding Semantic Matches Between Conceptual Graphs

Incorporating Rewrites

• With the introduction of rewrites, the match problem is redefined as:

Given:C1: A concept.C2: A concept.R: A set of rewrites.c: match criterion.

C1 and C2 semantically match iff by C1 * C1', C1' semantically matches C2

where r R.

r

Page 24: Finding Semantic Matches Between Conceptual Graphs

An Example

“A Man who blows up a trailer attached to the bumper of a car that he owns, which also has a chassis and a wheel, will cause the car to become detached.”

c: Exact match

Page 25: Finding Semantic Matches Between Conceptual Graphs

An Example: Intersection

Intersection of C1 and C2.

Page 26: Finding Semantic Matches Between Conceptual Graphs

An Example: Transitivity

Applying the Transitivity rule.

Page 27: Finding Semantic Matches Between Conceptual Graphs

An Example: Transitivity

Page 28: Finding Semantic Matches Between Conceptual Graphs

An Example: Part Ascension

Applying Part Ascension.

Page 29: Finding Semantic Matches Between Conceptual Graphs

An Example: Part Ascension

Page 30: Finding Semantic Matches Between Conceptual Graphs

An Example: Covers

defeated-by covers

caused-by

Page 31: Finding Semantic Matches Between Conceptual Graphs

An Example: Covers

Page 32: Finding Semantic Matches Between Conceptual Graphs

An Example: Match Completed

Intersection of C1 and C2 is not empty and c is satisfied

Page 33: Finding Semantic Matches Between Conceptual Graphs

Heuristic Rewrite Rules

• Heuristic rewrites differ from sound rewrites in only one way. They are not universally true.

• Whether or not they hold depends on the semantics of the things involved.

• Example:

Page 34: Finding Semantic Matches Between Conceptual Graphs

Pete’s Rudder Example

“The Pilot moved the rudder with the pedal.”

“The pilot pressed the pedal which causes the cable to be pull which in turn caused the rudder to move.”

c: Exact match.

Page 35: Finding Semantic Matches Between Conceptual Graphs

Pete’s Rudder Example

“The Pilot moved the rudder with the pedal.”

“The pilot pressed the pedal which causes the cable to be pull which in turn caused the rudder to move.”

Apply the rule:

Page 36: Finding Semantic Matches Between Conceptual Graphs

Pete’s Rudder Example

“The Pilot moved the rudder with the pedal.”

“The pilot pressed the pedal which causes the cable to be pull which in turn caused the rudder to move.”

Apply the rule again:

Page 37: Finding Semantic Matches Between Conceptual Graphs

Pete’s Rudder Example

“The Pilot moved the rudder with the pedal.”

“The pilot pressed the pedal which causes the cable to be pull which in turn caused the rudder to move.”

Assume additional information about the Cable and Pedal was defined.

Page 38: Finding Semantic Matches Between Conceptual Graphs

Pete’s Rudder Example

“The Pilot moved the rudder with the pedal.”

“The pilot pressed the pedal which causes the cable to be pull which in turn caused the rudder to move.”

Page 39: Finding Semantic Matches Between Conceptual Graphs

Pete’s Rudder Example

“The Pilot moved the rudder with the pedal.”

“The pilot pressed the pedal which causes the cable to be pull which in turn caused the rudder to move.”

Heuristic Rewrite: instrument covers is-part-of

Page 40: Finding Semantic Matches Between Conceptual Graphs

Pete’s Rudder Example

“The Pilot moved the rudder with the pedal.”

“The pilot pressed the pedal which causes the cable to be pull which in turn caused the rudder to move.”

Page 41: Finding Semantic Matches Between Conceptual Graphs

Pete’s Rudder Example

“The Pilot moved the rudder with the pedal.”

“The pilot pressed the pedal which causes the cable to be pull which in turn caused the rudder to move.”

Heuristic Rewrite: instrument covers has-part

Page 42: Finding Semantic Matches Between Conceptual Graphs

Pete’s Rudder Example

“The Pilot moved the rudder with the pedal.”

“The pilot pressed the pedal which causes the cable to be pull which in turn caused the rudder to move.”

Page 43: Finding Semantic Matches Between Conceptual Graphs

Pete’s Rudder Example

“The Pilot moved the rudder with the pedal.”

“The pilot pressed the pedal which causes the cable to be pull which in turn caused the rudder to move.”

Match between the input and the prior

Page 44: Finding Semantic Matches Between Conceptual Graphs

Talk Outline

• Motivation.

• Matching.

• Rewrite Rules.

• Matching in a KB.

• Elaboration.

• Applications.

• Future Work.

• Related Work.

Page 45: Finding Semantic Matches Between Conceptual Graphs

Matching in a KB

• In general, we are given a concept and an existing KB.

• Problem: Given a concept, find all the applicable concepts from the KB by applying the match test to each candidate.

• Formally,

Given: P: Prior Knowledge. I: Given concept.t: A minimum threshold.c: A match criterion.

Find: A subset P' P where for all p P', p and I semantically match and match-score(I, p) t.

Page 46: Finding Semantic Matches Between Conceptual Graphs

Controlling Search

• We must look through the KB to find the relevant concepts.

• This is very expensive.• Possible Solution: Index the prior knowledge in

some fashion so the entire KB does not need to be examined (work in progress).

• SMEs can help by: – selecting the most relevant piece of knowledge from a

set of matches.

– picking a starting point to search from.

– providing a set of candidates to match.

Page 47: Finding Semantic Matches Between Conceptual Graphs

Talk Outline

• Motivation.

• Matching.

• Rewrite Rules.

• Matching in a KB.

• Elaboration.

• Applications.

• Future Work.

• Related Work.

Page 48: Finding Semantic Matches Between Conceptual Graphs

Elaboration• Problem: Given a user concept and a relevant

prior, how can the two be overlaid s.t. the prior meaningfully elaborates the user concept.

• More specifically,

• We're aiming for a semi-automated approach to elaboration where the system suggests I' and the user can accept or modify I'.

Given: I: user graph p: An applicable prior knowledge

Generate: A new graph I' = I ° p.

Page 49: Finding Semantic Matches Between Conceptual Graphs

An Example of Elaboration

I:

p:

Definition of Conversion from the KB.

Description of Bioremediation entered by a SME.

Page 50: Finding Semantic Matches Between Conceptual Graphs

An Example of Elaboration

I’ = I ° p

Initial composition of bioremediation and conversion.

Page 51: Finding Semantic Matches Between Conceptual Graphs

An Example of Elaboration

I’ = I ° p

Page 52: Finding Semantic Matches Between Conceptual Graphs

Talk Outline

• Motivation.

• Matching.

• Rewrite Rules.

• Matching in a KB.

• Elaboration.

• Applications.

• Future Work.

• Related Work.

Page 53: Finding Semantic Matches Between Conceptual Graphs

Applications

• Semantic matching can be applied to a variety of applications:– Knowledge Acquisition.– Rule Bases in general.– Knowledge-based IR.– Question Answering.– Pattern Completion.

Page 54: Finding Semantic Matches Between Conceptual Graphs

Knowledge Acquisition

• Goal: To accelerate a SME's entry of knowledge by helping them locate applicable prior knowledge.

• Problem: – Existing KA tools do not reconcile new knowledge

with existing knowledge (Shaken).

– They do not identify relevant prior knowledge.

– SME has to be familiar with the KB in order to do knowledge entry effectively.

• Semantic matching can be used to locate relevant prior knowledge.

Page 55: Finding Semantic Matches Between Conceptual Graphs

Knowledge-Based IR

• Goal: To increase precision in information retrieval on digital libraries.

• Problem:– Statistical Methods rely on redundancy and co-

references in document.

– Existing approaches either do not fully exploit the KB or are limited w.r.t. the expressiveness of the query (McGuinness, Woods).

• Semantic matching addresses these issues and can be applied to this problem.

Page 56: Finding Semantic Matches Between Conceptual Graphs

Pattern Completion

• Problem: Given a user representation, elaborate it with a relevant piece of prior knowledge.

• This problem is useful for domains where speculation is needed (e.g. Battle Space Planning).

Page 57: Finding Semantic Matches Between Conceptual Graphs

An Example of Pattern Completion

I:

p:

c: Exact match.

Definition of Vertical-Envelopment from the KB.

SME: What is the larger context of this particular

landing?

Page 58: Finding Semantic Matches Between Conceptual Graphs

Pattern Completion: Definition Expansion

I:

Expand the definition of Flight

p:

Page 59: Finding Semantic Matches Between Conceptual Graphs

Pattern Completion: Transitivity

I:p:

Apply the transitivity rule.

Page 60: Finding Semantic Matches Between Conceptual Graphs

Pattern Completion: Transitivity

I: p:

Result of applying the transitivity rule.

Page 61: Finding Semantic Matches Between Conceptual Graphs

Pattern Completion: Rewrite Rule

Apply the rewrite:

Must first apply preparatory-event-of covers subevent to align LHS.

I:

p:

Page 62: Finding Semantic Matches Between Conceptual Graphs

Pattern Completion: Rewrite Rule

I: p:

The result of applying the previous rewrite.

Page 63: Finding Semantic Matches Between Conceptual Graphs

Pattern Completion: Covers and Heuristic Rule.

- Apply is-near covers location-of.

- Apply the heuristic rewrite that

I:

p:

allows siblings to match.

Page 64: Finding Semantic Matches Between Conceptual Graphs

Pattern Completion

Intersect of I and p with c also being satisfied.

I:p:

Page 65: Finding Semantic Matches Between Conceptual Graphs

Future Work• Identify more patterns to populate the library of

rewrites. • Identify types of discrepancies in representation

that rewrites can and cannot handle.• Identify the boundary of rewrites.• How to index prior knowledge so search can be

controlled?• How best to compose two concepts for

elaboration?• Apply this method to described applications and

verify utility through experimental studies.

Page 66: Finding Semantic Matches Between Conceptual Graphs

Related Work

• Conceptual Graphs (Sowa).• Matching

– Structure Mapping and Analogy (Forbus, Gentner, Markman).

– Literal Similarity (Tversky).

– Information Processing (Les Cohen).

– Graph Isomorphism.

– Subgraph morphism.

• Graph Transformations.