Bigscholar 2014, April 8, Seoul, South Korea - PowerPoint Presentation

Slide1

Bigscholar 2014, April 8, Seoul, South Korea

1

Trust and Hybrid Reasoning for Ontological Knowledge Bases

Hui

Shi, Kurt Maly, and Steven Zeil

Contact

: maly@cs.odu.eduSlide2

2

Outline

Problem

Semantic

web subject to changes

How to scale a reasoner to big data?Background Knowledge base using ontologiesInference strategiesBenchmarksOptimized backward chainingHybrid ReasonerMaterialization with search and retrieval Change and marking trusted/untrusted areasReason over untrusted goals Conservative trust assessmentProperty based trustPattern based trustEvaluationConclusions

Bigscholar

2014, April 8, Seoul, South KoreaSlide3

3

Problem

Efficiency of reasoning in the face of large scale and frequent changes within a question/answer system over a semantic web

Issue

Forward chaining scales well for fixed knowledge bases

Backward chaining can handle changes in knowledge base but does not scaleBigscholar 2014, April 8, Seoul, South KoreaSlide4

Background

Existing semantic application: question/answer systemsLibra,

Cimple

,

Arnetminer

Semantic WebResource Description Framework(RDF)Web Ontology Language (OWL) for specific knowledge domainsSPARQL query language for RDFSWRL rule languageReasoning systemsJena proprietary Jena rulesPellet and KANON ORACLE 11gOWLIMBigscholar 2014, April 8, Seoul, South Korea

4Slide5

5

Background

Knowledge base (KB)

Ontologies

Representation formalism: Description Logic (DL)

Inference methods for First Order LogicMaterialization and forward chaining pre-computes inferred truths and starts with the known data suitable for frequent computation of answers with data that are relatively staticOwlim and OracleQuery-rewriting and backward chaining expands the queries and starts with goals suitable for efficient computation of answers with data that are dynamic and infrequent queriesVirtuoso

Bigscholar 2014, April 8, Seoul, South KoreaSlide6

Background

Benchmarks evaluate and compare the performances of different reasoning systemsThe Lehigh University Benchmark (LUBM)

The University Ontology Benchmark (UOBM)

6

Bigscholar 2014, April 8, Seoul, South KoreaSlide7

Background

Optimized backward-chaining algorithmgenerate a query response for a given query pattern based on a specific rule set (RDFS

, Horst, custom)

Ordered

Selection Function

Switching between Binding Propagation and Free Variable Resolution Avoid Repetition and Non-Termination (OLDT)owl:sameAs Optimization Bigscholar 2014, April 8, Seoul, South Korea7Slide8

Hybrid reasoner

Motivation example

Assume fully materialized KB

Harvester adds new fact: student0 enrolled course0

Query ‘Who is enrolled in course 0?’ ok

Assume fact Porf0 teaches course0 in KBQuery “Who is being taught by Prof0?” not ok as simple lookup; needs reasoning with rule such as:enrolledIn(?Student,?Course?), teaches(?Faculty,?Course) :- isTaughtBy(?Student,?faculty)Bigscholar 2014, April 8, Seoul, South Korea8Slide9

Hybrid reasoner

Mark region of KB ‘trusted’ that is not affected by changeHybrid algorithm:

If a

goal

is in trusted region then return substitutions from KB

else for each rule R and substitution σ1 such that the head of R σ1 matches goal proveTheRuleBody (R.body, σ1)proveTheRuleBody: prove each goal in the rule body one by one recursively9Bigscholar 2014, April 8, Seoul, South KoreaSlide10

Trustworthy Goals

proof goal p(?X,?Y) is trustworthy if all instances of that goal derivable from facts and rules in the knowledge base are present in that knowledge base as instances

In practice, we will need to approximate set of trustworthy goals

A partition into trusted and untrusted sets is called

conservative

if no untrustworthy goals are trusted10Bigscholar 2014, April 8, Seoul, South KoreaSlide11

Approximation 1: Trusted Properties

property-based trust: assume that any property P that was involved in a change is itself untrusted

take

the closure of the “is used as a premise of” relation,

P

occurs in the body of a rule used to prove R …, P(x,y), … :- R(w,z) then R is also untrusted.Bigscholar 2014, April 8, Seoul, South Korea11Slide12

Approximation 1: Trusted Properties

property-based trust breaks down in the face of “meta-rules” in the knowledge base, rules that permit reasoning about properties

themselves, e.g., inverse rule

special handling of the meta-rules common to RDF and

OWL

result in significant fractions of the knowledge base being marked as untrusted unnecessarily12Bigscholar 2014, April 8, Seoul, South KoreaSlide13

Approximation 2: Trusted Patterns

Pattern-based trust: a pattern P(X,Y) (where X and Y could be ground instances or free variables) is

untrusted

if

it matches a change to the knowledge base or

if it can be derived from a rule with an untrusted pattern as a premiseOffers finer discrimination than property-based Bigscholar 2014, April 8, Seoul, South Korea13Slide14

Computing Untrusted Pattern

Marking algorithm

Add change

Check

each rule in the rule set to see if we can propagate the “

untrust” forward by a limited, specialized analogue of forward chainingAdd untrusted set produced from the above one change to the existing untrusted set, discarding any patterns that are specializations of other elements14Bigscholar 2014, April 8, Seoul, South KoreaSlide15

Untrusted Pattern

ExampleHarvester adds ‘worksFor(Fullprofessor0, University0)’

Marking algorithm discovers as untrusted

worksFor

(Fullprofessor0, University0)

member (University0, Fullprofessor0)memberOf (Fullprofessor0, University0)Query: “Who are members of University0?” needs reasoningQuery: “Who are members of University1” ok for direct retrieval as memberOf(?x, University1) is trustedBigscholar 2014, April 8, Seoul, South Korea15Slide16

Evaluation: property-based

16

Bigscholar 2014, April 8, Seoul, South Korea

Changes

Actual

# new properties

Actual

# new facts

#

untrusted properties

Adding a new class

2

3

12

Add a subclass relationship between two new classes

2

6

12

Add new Class as subClass of existing class

2

5

12

Adding a new Property

2

2

12

Add a new Property as subPropertyOf of another new Property

2

4

12

Add new Property as subPropertyOf of existing Property

2

3

12

Add new Class as domain to a new Property

3

5

13

Add new Class as range to a new Property

3

5

13Slide17

Evaluation: pattern-based

Produces the same number of properties as the ‘actual’ columns show

comparison of performance of our hybrid pattern-based proof algorithm against our regular, optimized backward chaining algorithm

and

against the OWLIM using LUBM1, LUBM10, and LUBM40, of size 100,839, 1,272,871, and 5,307,754

objects Query response time (ms) after adding student

17

Bigscholar 2014, April 8, Seoul, South KoreaSlide18

Evaluation

Query

response time (

ms

) after

adding undergraduate studentpercentage of untrusted facts in KB ranges 0 to a high of 10%percentage of untrusted patterns in KB ranges 0 to a high of 5%18

Bigscholar 2014, April 8, Seoul, South KoreaSlide19

19

Conclusions

We reported on our efforts to use ‘ trust’ in backward-chaining

reasoners

to accommodate the changing knowledge base.

We have shown that a pattern-based marking algorithm errs on the conservative side at an acceptable level and We show that compared to a forward chaining algorithm and a pure backward chaining algorithm that our hybrid algorithm is better in almost all cases testedBigscholar 2014, April 8, Seoul, South KoreaSlide20

Future Work

Explore the performance of the trust marking algorithm and of the hybrid reasoner

as a function of the fraction of the knowledge base that is

untrusted

Explore

the impact of long sequences of individual changes on the marking algorithm time and subsequently on the hybrid reasonerExplore performance of the hybrid reasoner as a function of the overall degree of inter-connection within the knowledge base semantics as a loosely connected network will lead to faster termination of the trust marking algorithm20Bigscholar 2014, April 8, Seoul, South Korea