Active Collaborations Istituto di Analisi dei Sistemi Informatici

Active Collaborations Istituto di Analisi dei Sistemi Informatici di Roma del CNR: Giorgio Gambosi, Carlo Gaibisso e Marco Biachi European Project Network Of Excellence - COREGRID: Antonia Ghiselli del INFN CNAF

Overview GRID SITEs and GRID COMPONENTs ONTOLOGY ●DESCRIPTION LOGIC FOR GRID-ONTOLOGY ●BINDING GRID-ONTOLOGY INTO ●RDFS and ●OWL-DL ●REASONING ON KNOWLEDGE BASE FOR ●MONITORING AND ●QUALITY OF SERVICE. ●

Declarative language and Logic System DECLARATIVE LANGUAGE blurs the distinction between a program as a set of instructions and a program as an ASSERTION ABOUT THE DESIRED ANSWER, LOGIC PROGRAM essentially is MATHEMATICAL ASSERTION. LOGIC SYSTEMS are Well Founded and based on SENTENCE CALCULUS. LOGIC SENTENCE CALCULUS is algebraic, so ASSERTIONS are/are verified as WELL FORMED EXPRESSIONS.

Declarative language and Logic System MAIN COMPONENTS ARE: RAW CONCEPTS ~ {JOB, HOST, MMEMORY, FILE, QUEUE, OID} AND AGGREGATION OF THEM ● TBOX ~ ASSERTIONS ON STRUCTURE WITH “SUBSUMPTION, BINARY RELATION, EQUIVALENCE” ● ● ABOX ~ ASSERTIONS ON INDIVIDUALS ● LBOX ~ THEOREMS AND AXSIOMS FOR REASONING

Declarative Language & Ontology DESCRIPTION LOGICS FOCUS ON ATTRIBUTES OF “WHAT” we want to know about some CONCEPTS (Asking to REASONER a QST about ADDRESSABLE(COMPUTING_ELEMENT) then ●Inferential Procedure let REASONER deducts some concept from concepts like ● ENOUGH_MEMORY_SPACE_LEFT, ●QUEUE_DEFINE and ●Relationships between them). PROCEDURAL LANGUAGEs and Systems based on them imply: Reasoning Activities are delegated to HUMAN REASONER who has to think about: “MEMORY_SPACE_LEFT is enough to QUEUE_DEFINITION that COMPUTING_ELEMENT is ADDRESSABLE ).

Description Logic Stratification

Grid Knowledge base example

Grid Knowledge base example I assert something valid in my INTERPRETING DOMAIN that is: CONSIDERING an HOST where is possible to ALLOCATE at least 1 MB of MMEMORY where it is possible to RESERVATE a FILE, that subsumes the existenc of a QUEUE THEN a JOB can be taken into count by the HOST. FORMALIZE THIS INTO A KB.

Grid Knowledge base example First REASONER classify CONCEPTS REASONER explicit the implicit information as CONCEPTS and RELATIONSHIPS. REASONER stores the THEOREMS and AXSIOMS REASONER wait for ASSERTIONS on CONCEPTS and REALATIONS RESAONER QUERY and DEDUCTION on Evolution of Grid-Componen Grid-Site life.

THE COUPLE < JOB, HOST > TRYING TO VERIFY SATISFIABILITY of <COUPLE OF CONCEPTS> GRID-SITES have different SEMANTIC for that couple. ANSWER can be some subsumed t-upla using some set of Theorems: ( < JOB(“OID”), HOST(“URL”) > ) |- < QUEUE, SUBMITABLE> ( < JOB(“OID”), HOST(“URL”) > ) |- < MEMORYSPACE, ALLOCABLE >

DEDUCTION ON COUPLE < JOB, HOST > FROM THAT ASSERTION ON GRID-ABOX <JOB, HOST> --> evaluate(NEW QUEUE) NEW QUEUE --> evaluate (BATCH SYSTEM POWER AND ALLOCATE CPU FOR SCHEDULER AND JOB WAITING LESS MMEMORY) THIS EVALUATION TAKE REASONER TO SIMULATE VARIOUS SCENARIO ON GRID-ABOX DEDUCTING SOMETHING FOR NEXT JOB REQUEST.

Autonomous GRID Well Defined Ontology of Grid-Sites, Grid-components, Grid-services Explicit Formalization with FOL of various kind of relationships among the previous CONCEPTS GIS will be an Inferential-GIS with ACTIVE, PRO-ACTIVE, DEDUCTIVE, QUERYING managing capacity. GIS will be able to use different Grid-TBOX at runtime.

Some Architectural Aspects

Open Aspects Formalization of GRID COMPONENT TBOX Formalization of GRID SITES TBOX Formalization of GRID THEOREM and GRID TAUTOLOGY OWL-DL of GRID KNOWLEDGE BASE INFERENTIAL PROCEDURE for GRID KNOWLEDGE BASE

References Reasoning with Individuals for the Description Logic SHIQ Ian Horrocks University of Manchester, Ulrike Sattler Lu. FG RWTH Aachen Stephan Tobies RWTH Aachen Tbox and Abox Reasoning in Expressive Description Logics Giuseppe de Giacomo La. Sapienza Univeristy, Maurizio Lenzerini La. Sapie University