Knowledge Representation and Reasoning in SNePS for Bioinformatics

Post on 20-Jan-2016

25 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

DESCRIPTION

Knowledge Representation and Reasoning in SNePS for Bioinformatics. Stuart C. Shapiro Department of Computer Science and Engineering, and Center for Cognitive Science University at Buffalo, The State University of New York 201 Bell Hall, Buffalo, NY 14260-2000 shapiro@cse.buffalo.edu - PowerPoint PPT Presentation

Transcript

cse@buff

alo

Knowledge Representationand Reasoning

in SNePSfor Bioinformatics

Stuart C. Shapiro Department of Computer Science and Engineering,

and Center for Cognitive Science

University at Buffalo, The State University of New York

201 Bell Hall, Buffalo, NY 14260-2000

shapiro@cse.buffalo.edu

http://www.cse.buffalo.edu/~shapiro/

http://www.cse.buffalo.edu/sneps/

November, 2003 S. C. Shapiro 2

cse@buff

alo

SNePS

A logic- and network-basedKnowledge representationReasoningand ActingSystem

November, 2003 S. C. Shapiro 3

cse@buff

alo

Examples of SNePSfor Bioinformatics

Neurological Diagnostic Expert SystemSystem for Understanding NF1 Literature

November, 2003 S. C. Shapiro 4

cse@buff

alo

NEUREX: A Neurological Diagnostic Expert System

Ca. 1983 – 1986See: Z. Xiang, J. G. Chutkow, S. C. Shapiro, and

S. N. Srihari, Computerized neurological diagnosis: a paradigm of modeling and reasoning, Health Care Instrumentation 1, 3 (1986), 90-105.

November, 2003 S. C. Shapiro 5

cse@buff

alo

Strengths of SNePSfor This Task

Integrated Analogical, Propositional, Functional Knowledge

Access to Procedural Knowledge via Acting System

November, 2003 S. C. Shapiro 6

cse@buff

alo

Analogical/PropositionalRepresentation

Proposition,

Ventral root is proximal to spinal nerve.

is represented by{<p “ventral root”> <d “spinal nerve”>}

also have{<p “ganglion”> <d “spinal nerve”>}

{<p “spinal nerve”> <d “dorsal ramus”>}

{<p “spinal nerve”> <d “ventral ramus-1”>}

November, 2003 S. C. Shapiro 7

cse@buff

alo Analogical/Propositional

From Z. Xiang, S. N. Srihari, S. C. Shapiro, and J. G. Chutkow, A modeling scheme for diagnosis, Expert Systems in Government Symposium, IEEE Computer Society Press, Silver Spring, MD, 1985, 538-547.

Propositions form a relational graph.

November, 2003 S. C. Shapiro 8

cse@buff

alo

A Knowledge-Based Approach to Understanding

the NF1 LiteratureA proposal to DoD

PI: Gary R. Skuse, Director of Bioinformatics, RIT CoPIs:

• Debra T. Burhans, Director, Bioinformatics, Canisius College

• Alistair E. R. Campbell, Computer Science, Hamilton College

• Stuart C. Shapiro, CSE, UB

November, 2003 S. C. Shapiro 9

cse@buff

alo

Goals

Discover new linkages in information across Medline abstracts of research literature

about Neurofibromatosis 1(NF1, or von Recklinghausen disease)

November, 2003 S. C. Shapiro 10

cse@buff

alo

Strengths of SNePSfor This Task

Inconsistency Tolerance/Discovery/RepairContextsMultiple SourcesAmenable to Backend DB Storage

November, 2003 S. C. Shapiro 11

cse@buff

alo

Inconsistent Knowledge Base

wff3: free(Willy) and whale(Willy)

wff10: all(x)(whale(x) => mammal(x))

wff19: all(x)(whale(x) => fish(x))

wff20: all(x)(andor(0,1){mammal(x), fish(x)})

wff21: all(x)(fish(x) <=> has(x,scales))

November, 2003 S. C. Shapiro 12

cse@buff

alo

Finding the ContradictionDuring Query Answering

: has(Willy, scales)?

I infer fish(Willy)

I infer it is not the case that

wff23: fish(Willy)

November, 2003 S. C. Shapiro 13

cse@buff

alo

BR AdviceIn order to make the context consistent you must delete

at least one hypothesis from the set listed below.

1 : wff20: all(x)(andor(0,1){mammal(x),fish(x)}) (1 dependent proposition: (wff24)) 2 : wff19: all(x)(whale(x) => fish(x))

(2 dependent propositions: (wff23 wff22)) 3 : wff10: all(x)(whale(x) => mammal(x))

(3 dependent propositions: (wff24 wff15 wff11)) 4 : wff3: free(Willy) and whale(Willy)

(8 dependent propositions: (wff24 wff23 wff22 wff11 wff9 wff5 wff2 wff1))

User deletes #2: wff19.

November, 2003 S. C. Shapiro 14

cse@buff

alo

Continuing with Repaired KB

I infer it is not the case that

wff22: has(Willy,scales)

wff26: ~has(Willy,scales)

November, 2003 S. C. Shapiro 15

cse@buff

alo

Multiple Sourceswff1: all(x)(andor(0,1){mammal(x),fish(x)})

wff2: all(x)(fish(x) <=> has(x,scales))

wff4: all(x)(whale(x) => fish(x))

wff5: Source(Melville,all(x)(whale(x) => fish(x)))

wff6: all(x)(whale(x) => mammal(x))

wff7: Source(Darwin,all(x)(whale(x) => mammal(x)))

wff8: Sgreater(Darwin,Melville)

wff11: free(Willy) and whale(Willy)

Note: Source & Sgreater props are regular object-language props.

November, 2003 S. C. Shapiro 16

cse@buff

alo

: has(Willy, scales)?

I infer fish(Willy)

I infer has(Willy,scales)

I infer mammal(Willy)

I infer it is not the case that wff14: fish(Willy)

Finding the Contradiction

November, 2003 S. C. Shapiro 17

cse@buff

alo

Using Source CredibilityA contradiction was detected within context default-defaultct.The contradiction involves the newly derived proposition:

wff17: ~fish(Willy) {<der,{wff1,wff6,wff11}>}

and the previously existing proposition: wff14: fish(Willy) {<der,{wff4,wff11}>}

The least believed hypothesis: (wff4) The most common hypothesis: (nil) The hypothesis supporting the fewest wffs: (wff1)

I removed the following belief: wff4: all(x)(whale(x) => fish(x))

I no longer believe the following 2 propositions: wff14: fish(Willy) wff13: has(Willy,scales)

November, 2003 S. C. Shapiro 18

cse@buff

alo

Backend DB Storage

A proposalAll propositions represented with keyword

argumentsEach keyword set a Relation in an RDBOr in XML

November, 2003 S. C. Shapiro 19

cse@buff

aloNEUREX Data as RelationP Proximal Distalm1 dorsal root ganglionm2 ganglion spinal nervem3 ventral root spinal nervem4 spinal nerve dorsal ramusm5 dorsal ramus medial branch of d.r.m6 medial branch of d.r. medial cutaneous branchm7 medial branch of d.r. lateral cutaneous branchm8 dorsal ramus lateral branch of d.r.m9 spinal nerve ventral ramus-1m10 ventral ramus-1 ventral ramus-2m11 ventral ramus-1 lateral cutaneous branch of v.r.m12 cutaneous branch of v.r. posterior branch lateralm13 cutaneous branch of v.r. anterior branch lateral

November, 2003 S. C. Shapiro 20

cse@buff

alo

Reasoning in Different Contexts

A context is a set of hypotheses and all propositions derived from them.

Reasoning is performed within a context.A conclusion is available in every context that

is a superset of its origin set.Contradictions across contexts are not noticed.

November, 2003 S. C. Shapiro 21

cse@buff

alo

Darwin Context

: set-context Darwin ()

: set-default-context Darwin

wff1: all(x)(andor(0,1){mammal(x),fish(x)})

wff2: all(x)(fish(x) <=> has(x,scales))

wff3: all(x)(orca(x) => whale(x))

wff4: all(x)(whale(x) => mammal(x))

wff7: free(Willy) and whale(Willy)

November, 2003 S. C. Shapiro 22

cse@buff

alo

Melville Context

: set-context Melville (wff8 wff7 wff3 wff2 wff1)

: set-default-context Melville

wff9: all(x)(whale(x) => fish(x))

November, 2003 S. C. Shapiro 23

cse@buff

alo

Melville: Willy has scales: has(Willy, scales)?

I infer fish(Willy)

I infer has(Willy,scales)

wff10: has(Willy,scales)

November, 2003 S. C. Shapiro 24

cse@buff

alo

Darwin: No scales: set-default-context Darwin

: has(Willy, scales)?

I infer mammal(Willy)I infer it is not the case that wff11: fish(Willy)

I infer it is not the case that

wff10: has(Willy,scales)

wff15: ~has(Willy,scales)

November, 2003 S. C. Shapiro 25

cse@buff

alo

Summary

SNePS is useful for Bioinformatics reasoning.Analogical/Propositional/DB Representation.Notifies user about inconsistencies.Resolves inconsistencies from differently

credible sources.Can reason in one context, even if another is

inconsistent with it.

top related