a translator framework for the interoperation of graph...
Post on 06-Feb-2018
216 Views
Preview:
TRANSCRIPT
IntroductionCombined Interoperation/Portability Methods
Use Cases
A Translator Framework for the Interoperationof Graph plus Relational Data and Rules
on the WebAWoSS 2014
Saint John, NB, Canada, February 24, 2014
Gen Zou, Harold Boley
Faculty of Computer Science,University of New Brunswick, Fredericton, Canada
1 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Outline
1 Introduction
2 Combined Interoperation/Portability MethodsObject-Relational Interoperation FrameworkPortability ArchitectureEvaluation
3 Use Cases
2 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Outline
1 Introduction
2 Combined Interoperation/Portability MethodsObject-Relational Interoperation FrameworkPortability ArchitectureEvaluation
3 Use Cases
3 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Rule Languages
Along with ontologies, rules provide foundation ofknowledge representation and problem solvingUsed to express
Knowledge for ontology/rule-based data accessDomain-specific (e.g., biomedical) concept definitionsAssociations among dataBusiness logicsPrivacy/security/trust policiesLegal norms...
Will enable Semantic Analytics of Big Data
4 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Rule Languages
Paradigms for modeling entity dependencies:RelationalGraph (Object-Centered)Combined
Since Knowledge Bases (KBs) have been developed inlanguages following all three paradigms, cross-paradigmtranslation, integration, and reuse is often necessaryNeed for an interoperation language and technology:Positional-Slotted Object-Applicative (PSOA) RuleMLNaturally combinable with portability technology:Platform-independent implementation of PSOA RuleML
5 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Hypergraph Example – Relational Betweenness
betweenRel
canadapacific atlantic
usa
mexico
Directed hyperarcs cut through intermediate nodes (cf. Grailog)
FactsbetweenRel(pacific, canada, atlantic)betweenRel(canada, usa,mexico)
6 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Graph Example – Object-Centered Betweenness
betweenObj
b0
b1
canadapacific atlantic
usa
mexico
outer1
inner
outer2
outer1
inner
outer2
Facts“#” denotes “∈” for class membership; “→” associates a slot name with its filler
b0#betweenObj(outer1 → pacific; inner → canada; outer2 → atlantic)b1#betweenObj(outer1 → canada; inner → usa; outer2 → mexico)
7 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Example – Integrated Betweenness (Enriched)
betweenObjRel
b0
b1
2
canadapacific atlantic
usa
mexico
westEast
northSouth
dim
orient
dim
orient
Factsb0#betweenObjRel(pacific, canada, atlantic; dim → 2; orient → westEast)b1#betweenObjRel(canada, usa,mexico; dim → 2; orient → northsouth)
8 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Relational Rule Languages
Widely used for relational DBs (SQL views) and KBs,representing information in classical logicModel dependencies among n entities as an n-arypredicate applied to an ordered sequence of n arguments,called positional argumentsLanguages: Common Logic, Prolog, TPTP-FOF, ...
9 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Graph (Object-Centered) Rule Languages
Receive increasing attention because of expandingresearch and development in linked data on the Web,graph/‘triple’ stores, and big data in NoSQL DBsEach object is represented by a unique Object IDentifier(OID), typed by a class, and described by an unorderedcollection of slots, each being a pair of a name and a fillerAn OID-describing slotted term in AI is called a frame(represents a resource/‘subject’-describing property list onthe Semantic Web)Languages: RDF, N3, ...
10 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Object-Relational Rule Languages
Combine the object-centered and relational paradigms,either in a heterogeneous or a homogeneous wayHeterogeneous
Allow atomic formulas in both relational and object-centeredforms, even mixed in the same ruleLanguages: F-logic and RIF
HomogeneousIntegrate relational and object-centered atomic formulasinto a unified formLanguage: PSOA RuleML
11 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
PSOA RuleML
Integrates relational and object-centered modelingGeneralizes F-logic, RIF-BLD, and POSLUses positional-slotted object-applicative (psoa) terms,permitting a relation application to have an OID – typed bythe relation – and, orthogonally, its arguments to bepositional or slotted.General case (multi-tuple):o#f([t1,1 ... t1,n1] ... [tm,1 ... tm,nm] p1->v1 ... pk->vk)
Special cases (single-tuple brackets and zero-argumentparentheses optional):Combined: o#f([t1 ... tn] p1->v1 ... pk->vk)Positional: o#f([t1 ... tn])Slotted: o#f( p1->v1 ... pk->vk)Member-only: o#f()
12 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Example of Querying a PSOA Fact and Rule
KB:b1#betweenObjRel(canada usa mexico
dim->2 orient->northSouth)
Forall ?out1 ?in ?out2 ?b(
?in#GeoUnit(neighborNorth->?out1neighborSouth->?out2) :-
?b#betweenObjRel(?out1 ?in ?out2orient->northSouth)
)
English Query: “Which GeoUnit has Canada as its northern neighbor?”Query: ?X#GeoUnit(neighborNorth->canada)Answer: ?X=usa
TPTP version in Common Logic would contribute to COLORE
13 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Semantics-Preserving Translation
For a translation ϕL1,L2 from the source language L1 to thetarget language L2:
Sound: all entailments that hold after translation to L2already hold in L1Complete: all entailments in L1still hold after translation to L2Semantics preserving = sound + completeSemantics-preserving translation desired forrealizing L1 via L2:
KB1
Q1
KB2
Q2
`1 `2
ϕL1,L2
ϕL1,L2
14 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Object-Relational Interoperation FrameworkPortability ArchitectureEvaluation
Outline
1 Introduction
2 Combined Interoperation/Portability MethodsObject-Relational Interoperation FrameworkPortability ArchitectureEvaluation
3 Use Cases
15 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Object-Relational Interoperation FrameworkPortability ArchitectureEvaluation
Outline
1 Introduction
2 Combined Interoperation/Portability MethodsObject-Relational Interoperation FrameworkPortability ArchitectureEvaluation
3 Use Cases
16 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Object-Relational Interoperation FrameworkPortability ArchitectureEvaluation
System of Translation Pathways
Use the homogeneous object-relational languagePSOA RuleML as the interchange languageImplement unidirectional or bidirectional translationsbetween PSOA RuleML and
Relational languages: TPTP and the Horn subset of PrologObject-centered language: N3Heterogenous object-relational language: RIF
17 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Object-Relational Interoperation FrameworkPortability ArchitectureEvaluation
General Translation Steps
Translations with PSOA RuleML as the source languageTransform the source KB into a normalized form with onlyelementary constructs, using a composition oftransformations staying within PSOA RuleML
ObjectificationSlotribution/tupributionSkolemizationUnnesting
Map elementary constructs into the target languageTranslations with PSOA RuleML as the target language
Mostly syntactic translation, target different subsets ofPSOA RuleML
18 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Object-Relational Interoperation FrameworkPortability ArchitectureEvaluation
General Translation Steps
Each translation is a recursive algorithm τ that can bespecified as a mapping table
Example of τpsoa: Mapping from PSOA RuleML to TPTP
PSOA/PS Constructs TPTP Constructso#f() member(τpsoa(o), τpsoa(f))
AND(f1 ... fn) (τpsoa(f1) & ... & τpsoa(fn))ϕ :- ψ τpsoa(ψ)=> τpsoa(ϕ)
... ...
19 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Object-Relational Interoperation FrameworkPortability ArchitectureEvaluation
Outline
1 Introduction
2 Combined Interoperation/Portability MethodsObject-Relational Interoperation FrameworkPortability ArchitectureEvaluation
3 Use Cases
20 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Object-Relational Interoperation FrameworkPortability ArchitectureEvaluation
Portability Architecture
Realizing increasing subsets of PSOA RuleML semanticsin portablility architecture: Framework for translators andtranslator-based reasoning systems
Instantiated in ANTLR (ANother Tool for LanguageRecognition) for executable translators implementingmapping tables, e.g. targeting L2 = TPTP
21 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Object-Relational Interoperation FrameworkPortability ArchitectureEvaluation
Portability Architecture
ANTLR-based Java library of translatorsParsers/Generators: between concrete-syntax strings andabstract-syntax trees (ASTs)Transformers: from ASTs to ASTs
Service-based composition technology for translatorsalong the pathways
22 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Object-Relational Interoperation FrameworkPortability ArchitectureEvaluation
Outline
1 Introduction
2 Combined Interoperation/Portability MethodsObject-Relational Interoperation FrameworkPortability ArchitectureEvaluation
3 Use Cases
23 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Object-Relational Interoperation FrameworkPortability ArchitectureEvaluation
Evaluation
Evaluation metrics for soundness and completeness,analogous to, respectively, precision and recall ininformation retrievalTest Suite
Contains existing independent test cases, e.g. fromhttp://www.w3.org/2005/rules/wiki/Category:Test_Case
and test cases developed by ourselves, athttp://wiki.ruleml.org/index.php/PSOA_RuleML
Also contains new use cases (following slides)Covers different language and translator features
24 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Outline
1 Introduction
2 Combined Interoperation/Portability MethodsObject-Relational Interoperation FrameworkPortability ArchitectureEvaluation
3 Use Cases
25 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Ongoing: Geospatial Rules
Extend our running example by formalizing and integratingfurther geospatial knowledgeFor each relation/class, determine most suitable paradigm(relational-only, object-centered-only, or object-relationalcombined) along with its argument treatmentEnhance geospatial data with knowledge in a Rule-BasedData Access (RBDA) scenario, e.g. for WSL forestry data
26 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Planned: Financial Business Rules
Develop KB with business rules in PSOA RuleML for thefinancial management of organizations, e.g. RuleML Inc.Data from financial documents and electronic financialstatement spreadsheets will be mapped to PSOA factsOIDs are used to connect metadata (KB) and paperdocuments. Can also be used as dereferenceable URLsfor scanned versions of documents.A systematics of financial business rules will be explored(e.g., validation rules for transaction plausibility checks,transformation rules for currency conversion, heuristicrules for checking-to-savings transfers)
27 / 28
IntroductionCombined Interoperation/Portability Methods
Use Cases
Demo
PSOATransRun available online:http://psoa.ruleml.cloudbees.net
Try test cases and send feedbackhttp://wiki.ruleml.org/index.php/PSOA_RuleML
28 / 28
top related