260 likes | 268 Views
Lecture 4: Handling Heterogeneity with Information Resource Dictionary Systems. Nick Rossiter, Computing Science, Newcastle University, England B.N.Rossiter@newcastle.ac.uk http://www.cs.ncl.ac.uk/people/b.n.rossiter/. Introduction. Looking at multi-level architecture for information systems
E N D
Lecture 4: Handling Heterogeneity with Information Resource Dictionary Systems Nick Rossiter, Computing Science, Newcastle University, England B.N.Rossiter@newcastle.ac.uk http://www.cs.ncl.ac.uk/people/b.n.rossiter/
Introduction • Looking at multi-level architecture for information systems • Need to handle higher levels than intension • Take object work two levels higher • Example with dates • Composition of functors and natural transformations
Approach • Principles of Information Resource Dictionary System (IRDS) • Benefits in General • Assistance for Cross-Platform Operation • Search for Sound Theoretical Basis
IRDS/Reference Model • Based on ISO/IEC Standards • 10027 1990 Information Res. Dict. Sys. (IRDS) • 10032 1993 Reference Model • 10728 1993 IRDS - System Interfaces; • 9075 1992 SQL
Outline IRDS • Introduces concept of meta-meta data • Four-level approach
Example of Dates • IRDDS concepts is some measure of time • IRDD is constructs available -- light years, years since various events, months, days, carbon dates, etc • IRD is formats available -- yyyy, mm/dd/yyyy, dd/mm/yyyy, etc • APP is values -- 195, 05/30/1967, 30/05/1967
Relating Extension to Intension levelexample • Application data Supplier(1,’Smith’) extension -direct addressing • Schema data Supplier (supp_no, name) intension - addressing by name level pair:
Relating Intension to Constructs • Schema data Supplier (supp_no, name) intension - addressing by name • Meta data Table(supplier) data dictionary - addressing by concepts, constructs.
Relating Constructs to Abstractions • Meta data Table(Supplier) data dictionary - addressing by concepts, constructs. • Meta meta data Aggregation(Table) real-world abstractions - addressing by abstractions
Level-pairs are the Critical Dictionary Entries • Meta is: mapping from schema names (intension) to constructs available as 2-tuple: <intension_entry, construction> e.g. <student<id,name>, table> • MetaMeta is: mapping from construct available to abstraction as 2-tuple: <construction, abstraction> e.g. <table, aggregation>
Examples of Levels and Level-pairs: Intension/Extension • Extension (APP): Student<1234, ‘John James’, ‘15 Montford Road’> • Intension (IRD): Student<id, name, address> • APP IRD by: Name: Student<1234, ‘John James’, ‘15 Montford Road’> Student<id, name, address>
Examples of Levels and Level-pairs: Constructs/Intension • Intension (IRD): Student<id, name, address> • Constructs (IRDD): Table • IRD IRDD by Meta: Student<id, name, address> Table
Examples of Levels and Level-pairs: Abstractions/Constructs • Constructs (IRDD): Table • Abstractions (IRDDS): Aggregation • IRDD IRDDS by MetaMeta: Table Aggregation
Interpretation of IRDS in Schematic form • concepts mission • MetaMeta Policy • constructs management • Meta Organize • intension enterprise • Name Data • extension information Information Resource Dictionary Definition Schema (IRDDS) Information Resource Dictionary Definition (IRDD) Information Resource Dictionary (IRD) Information Resource Data (APP)
Overall Mappings • Can write: System = MetaMeta o Meta o Name (APP) to represent composition. • System(APP) = IRDDS • Also the dual: Model = Data o Organize o Policy
IRDS Enables Heterogeneous Systems to be Related • Only one IRDDS in principle but extensible IRDDS covers all paradigms • For relational database, can use mapping: System : APP IRDDS • Comparison of System for each paradigm enables in principle: • completely different approaches to be related • heterogeneous systems to be run in integrated fashion
Formalization of IRDS Approach • Quality enhanced by formalization • Can know more confidently the limits and behaviour of a system • Which formalism is best-suited?
Category Theory? • Need relations to represent level-pairs • Multi-level construction (higher-order) • Category theory seems appropriate: - categories for basic structures - functors relate categories - natural transformations relate functors
IRDS Levels in Functorial Terms MetaMeta • IRDDSIRDD SystemPolicy Model Meta Data Organize • APPIRD Name
Dates Revisited Concept (IRDDS) date Policy MetaMeta Constructs (IRDD) giga years days,months,years (years days, months, years (years AD) Muslim) Organize Universe American Euroopean Muslim Meta Format (IRD) g.f x 109 mm/dd/yyyy dd/mm/yyyy dd/mm/yyyy Data Name Values 4.2 05/19/2000 19/05/2000 14/03/1350 (APP)
Connections between levels are Adjunctions • For two categories A and B, an adjunction between them involves two functors: • a free functor F: A B adding structure to A • an underlying functor G: B A forgetting structure in B • Organize: IRDD IRD and • Meta: IRD IRDD are a pair of adjunctions
Date Comparison as Natural Transformation • Take composition of adjuntions: Model_US: date 05/19/00 Model_Eur: date 19/05/00 Then have natural transformation: Amer_Eur_Date_Compare: Model_US Model_Eur
Further Reading • Information technology - Information Resource Dictionary System (IRDS) framework, Standard ISO/IEC 10027 (1990); 10728 (1993). • Constructing Standards for Cross-Platform Operation, Heather, M A, & Rossiter, B N, Constructing Standards for Cross-Platform Operation, Software Quality Management VI, British Computer Society, Amsterdam, April 1998, 218-229 (1998).
Concluding Remarks • Lecture 1 -- Interoperability -- showed problems of handling heterogeneous data • Lecture 2 -- Introduction to Category Theory -- described categories as a basis for thinking about information systems • Lecture 3 -- Objects as Categories -- described how the intension/extension could be represented by categories, functors and natural transformations
Finale • Lecture 4 -- IRDS -- showed how all four levels of an information system can be described formally with categories, perhaps giving a better understanding of problems.