ࡱ; (  !"#$%&'*+,-./0123456789:;<=>?@ABDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~Root Entry  ®`VTextStarWriter 5.0}`m|`|azɲa/ִώRnMtS@o:8y;" n̯-X .#F$R:T\Σ^1u-ʁSfxDocumentInfo  Graham Klyne S11 Graham Klyne AU1F  Graham Klyne AU1  Info 0 Info 1 Info 2 Info 3 S11VN"<44Standard LIBIMBEDDED LIBIMBEDDED TASK,0,1,H-2,0,100,1,5024;748;100;570;568;15225;8728;0;0SBX sb Z Standard StarBASICSBX ARSBX AR SBX AR2c%bqqOh+'0 h t 34@hLY@0Y@SW5HDR.0AU1F !{Numbering SymbolsBullet Symbols4Strong Emphasis5 Source Text Superscript3Citation8Variable Internet link9 Definition2EmphasisDefault Paragraph FontNormalVisited Internet Link Line numberingList 1 294856481 RTF_Num 2 373160971 373340501 373428231 373445421 373504971 373529031 373640271Outline0 #R  n*StarBats n*StarBats n*StarBats n*StarBats n*StarBats` n*StarBats@ n*StarBats  n*StarBats n*StarBats n*StarBats R  n).Symbol n).Symbol6 n).SymbolQ n).Symboll n).Symbol n).Symbol n).Symbol n).Symbol n).Symbol  n).Symbol  RW n#.0n#. 0n#.p0n#) @ 0n$()0n$() 0n$()0n$() 0n$()P0R  n).Symbol n+.StarBats6 n+.StarBatsQ n+.StarBatsl n+.StarBats n+.StarBats n+.StarBats n+.StarBats n+.StarBats n+.StarBats R  n).Symbol n+.StarBats6 n+.StarBatsQ n+.StarBatsl n+.StarBats n+.StarBats n+.StarBats n+.StarBats n+.StarBats n+.StarBats R  n).Symbol n+.StarBats6 n+.StarBatsQ n+.StarBatsl n+.StarBats n+.StarBats n+.StarBats n+.StarBats n+.StarBats n+.StarBats R  n).Symbol n+.StarBats6 n+.StarBatsQ n+.StarBatsl n+.StarBats n+.StarBats n+.StarBats n+.StarBats n+.StarBats n+.StarBats R  n).Symbol n+.StarBats n+.StarBatsI n+.StarBats n+.StarBats n+.StarBats n+.StarBatsU n+.StarBats n+.StarBats n+.StarBats R  n).Symbol n+.StarBats6 n+.StarBatsQ n+.StarBatsl n+.StarBats n+.StarBats n+.StarBats n+.StarBats n+.StarBats n+.StarBats R  n).Symbol n+.StarBats6 n+.StarBatsQ n+.StarBatsl n+.StarBats n+.StarBats n+.StarBats n+.StarBats n+.StarBats n+.StarBats Ze3Ni6D*z˃FՂ a6@d/*[Y*ptI/ O Rs0c3KY&1D rUk8@Cn0}Y`<+*er8RCF -E:GTr*ɺ^~9s.C'$τD :@*(\±z)l6aLFW++L$!&LSI3@)x"DSAQ|AMsI93SebHq:^KtU9ؘ^Vj d07=JuՂ SWG, A<  #$%&'()*./0123456789:;<=>?@ABCDGHK  !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFQRSTU0!':@XXX&%'@:eX'@> StarBats Courier NewTimes New RomanArialArialRegularXX!X7XQXa' @ d d  d d K d @d hd xd XXX"X,X6X@XJXTX^$' @ X' @&XX'@IX'@mXX1'#@X+'$(@XXX7'%@X8'&@ XX9''@5XX'( @ n. . . 8. q . q ..V..;.. ....XXX&X2XEXQX]XiXu@'1@rdrddrdddhdddhddddd d ddd#ddddddddd 6ddd6 QdddQ ldddl ddd 7ddd77d7dd7lddddddQdddl6dddQddd6ddd@dddhXX!X4XGXZXmXXXXXXXXXX+X>XQXdXwXXA'2@ dd ddhddddddd dddd<dxdxd  dd dxd dd dxd  d dddddXXX&X2X>XJXVXbXnXzXXXXX8S<@ X('@$@PXR'D@oXJJ@XTu ,Q *P $' '(. . p. @ . . . . . P. . . !. $. `'. 0*. -. /. 2. p5. @8. ;. =. @. 6')$2$ y,Numbering SymbolsNumbering SymbolsBullet SymbolsBullet Symbols''Strong EmphasisStrong Emphasis'4 Source Text Source Text@''5 Superscript Superscript%'CitationCitation@ '3VariableVariable '8 Internet link Internet link@!'' Definition Definition@ ''9EmphasisEmphasis@ '2Default Paragraph FontNormalDefault Paragraph Font'' $'NormalNormal'' $'Visited Internet LinkVisited Internet Link@!''Line numberingLine numbering FormulaFormula1@'<8S>> StandardStandard@$+'&8''9'('2A'33 Text bodyStandard Text body@'#1'&8''9'1@'2A'33Text body indent Text bodyText body indent1@' SignatureStandard SignatureJJ SalutationStandard SalutationJJ Marginalia Text body Marginalia1@' List Indent Text body List Indent('1@'HeadingStandard Text body@''$+'2A'33DR'  Heading 9Heading Text body'' Heading 8Heading Text body'' Heading 7Heading Text body'' Heading 6Heading Text body'' Heading 5Heading Text body' '' Heading 4Heading Text body' ''  Heading 3Heading Text body@''2A'33  Heading 2Heading Text body@'2A'33  Heading 10Heading Text body'' Heading 1Heading Text body@''2A' Hanging indent Text bodyHanging indent('1@'First line indent Text bodyFirst line indent1@'TitleHeadingSubtitle@''$+'2A'(SubtitleHeading Text body@' ''$+'2A'(AbstractStandardAbstract#1'1@'Heading prefaceHeading Text body'HeaderStandardHeader@''('JJAuthor detailsHeadingAuthor details '$+'2A'KeywordsStandardKeywords'Contents HeadingHeadingContents Heading@''1@'JJ IndexStandardIndex@'2A'JJ Contents 1Index Contents 1@('1@'2A' Contents 2Index Contents 2@('1@' Contents 3Index Contents 3@('1@'  Contents 4Index Contents 4('1@'  Contents 5Index Contents 5('1@'  AddresseeStandard Addressee2A'JJCaptionStandardCaption' '2A'JJEndnoteStandardEndnote'1@' JJFooterStandardFooter@'('JJ Footer leftStandard Footer left('JJ Footer rightStandard Footer right('JJFootnoteStandardFootnote'1@' JJ  Header leftStandard Header left('JJ Header rightStandard Header right('JJHorizontal LineStandard Text body'2A' @('JJ0 List ContentsStandard List Contents1@' 0 List HeadingStandard List Contents1@'0Preformatted TextStandardPreformatted text continued@''$+'1@'2A'330 QuotationsStandard Quotations1@'2A' 0SenderStandardSender2A'JJFrame contents Text bodyFrame contents List Text bodyListTable Contents Text bodyTable ContentsJJ Table HeadingTable Contents Table Heading ''$+'JJNumbering 5 StartListNumbering 5 Start1@'2A' Numbering 5 EndListNumbering 5 End1@'2A' Numbering 5 Cont.ListNumbering 5 Cont.1@'2A'  Numbering 5List Numbering 51@'2A' Numbering 4 StartListNumbering 4 Start1@'2A'  Numbering 4 EndListNumbering 4 End1@'2A' Numbering 4 Cont.ListNumbering 4 Cont.1@' 2A'  Numbering 4List Numbering 41@'2A' Numbering 3 StartListNumbering 3 Start1@'2A'  Numbering 3 EndListNumbering 3 End1@'2A'  Numbering 3 Cont.ListNumbering 3 Cont.1@' 2A'   Numbering 3List Numbering 31@'2A'  Numbering 2 StartListNumbering 2 Start1@'2A' Numbering 2 EndListNumbering 2 End1@'2A' Numbering 2 Cont.ListNumbering 2 Cont.1@' 2A'  Numbering 2List Numbering 21@'2A' Numbering 1 StartListNumbering 1 Start1@' 2A' Numbering 1 EndListNumbering 1 End1@' 2A' Numbering 1 Cont.ListNumbering 1 Cont.1@'2A'  Numbering 1List Numbering 11@' 2A'  List 5 StartList List 5 Start1@'2A' % List 5 EndList List 5 End1@'2A' ' List 5 Cont.List List 5 Cont.1@'2A' (List 5ListList 51@'2A' & List 4 StartList List 4 Start1@'2A' ! List 4 EndList List 4 End1@'2A' # List 4 Cont.List List 4 Cont.1@' 2A' $List 4ListList 41@'2A' " List 3 StartList List 3 Start1@'2A'  List 3 EndList List 3 End1@'2A'  List 3 Cont.List List 3 Cont.1@' 2A'  List 3ListList 31@'2A'  List 2 StartList List 2 Start1@'2A'  List 2 EndList List 2 End1@'2A'  List 2 Cont.List List 2 Cont.1@' 2A' List 2ListList 21@'2A'  List 1 StartList List 1 Start1@' 2A'  List 1 EndList List 1 End1@' 2A'  List 1 Cont.List List 1 Cont.1@'2A' List 1ListList 11@' 2A'  Superscript Superscript%'('33Preformatted text continuedPreformatted TextPreformatted text continued&8''9'2A'33 ReferenceStandard Reference%7'&8''9'('1@'33Text body blank before Text body Text body2A' Preformatted text hangingPreformatted TextPreformatted text continued$+'('1@'2A'33Text body.hanging-indent Text bodyText body.hanging-indent1@'Note Text bodyNote1@'TableCaptionTable Object index headingHeadingObject index heading''1@'JJ Object index 1IndexObject index 1('1@' NormalStandardNormal'' $' heading 3NormalNormal'$+'DR' Normal (Web)Normal Normal (Web)!''' $'2A'NP:n `;3s`)lV5  q @ , n = S%oI*-pF2vbB&rV6|> FNV$^5p !1!!!%"j"t0;QK:I.BoJe3"g7/i:/f%$J_g_obN%/apd cr{6h c/1R)N$'1 / @7"⃪lrr+w#2L.qc\ /q[{D/"h\*N/*0_O?!;kPREr`vDd&dX&Q%~$*`K'4]ҁ,cpʥ1F2#Pǐ[dI :LJH҅~fɾ66rmP%.388En@ىH Graham Klyne Graham KlyneSW5HDR.0AU1F !9 Frameformat ZeichenformatTextformatvorlageStandard Text bodyText body indent Signature Salutation Marginalia List Indent Heading Heading 9 Heading 8 Heading 7 Heading 6 Heading 5  Heading 4  Heading 3  Heading 2 Heading 10  Heading 1Hanging indentFirst line indent(Title(SubtitleAbstractHeading prefaceHeaderAuthor detailsKeywords Contents Heading Index Contents 1 Contents 2 Contents 3 Contents 4 Contents 5 AddresseeCaptionEndnoteFooter Footer left Footer right Footnote Header left Header right0Horizontal Line0 List Contents0 List Heading0Preformatted Text0 QuotationsSender Frame contentsListTable Contents Table HeadingNumbering 5 StartNumbering 5 EndNumbering 5 Cont. Numbering 5 Numbering 4 StartNumbering 4 EndNumbering 4 Cont. Numbering 4 Numbering 3 Start Numbering 3 End Numbering 3 Cont.  Numbering 3Numbering 2 StartNumbering 2 EndNumbering 2 Cont. Numbering 2Numbering 1 StartNumbering 1 EndNumbering 1 Cont. Numbering 1% List 5 Start' List 5 End( List 5 Cont.&List 5! List 4 Start# List 4 End$ List 4 Cont."List 4 List 3 Start List 3 End  List 3 Cont.List 3 List 2 Start List 2 End List 2 Cont.List 2 List 1 Start List 1 End List 1 Cont.List 1 SuperscriptPreformatted text continued ReferenceText body blank beforePreformatted text hangingText body.hanging-indentNote Table Object index heading Object index 1Normal heading 3 Normal (Web)Numbering SymbolsBullet Symbols4Strong Emphasis5 Source Text Superscript3Citation8Variable Internet link9 Definition2EmphasisDefault Paragraph FontVisited Internet Link Line numbering FormulaRoot 124 125 126Standard First Page  Illustration TextDrawing Contents 6 Contents 7 Contents 8 Contents 9 Contents 10Y .Y g.Y .Y .q'& GeneralGeneraldNC#,##0.00#,##0.00SystemNC #,##0.00 CCC#,##0.00 CCCNC$#,##0.--;[RED]($#,##0.--) $#,##0.--($#,##0.--)REDNC$ MM/DD/YYYYMM/DD/YYYY def/SystemNC%MM/DD/YYMM/DD/YY def/SystemNC&NNNNMMMM DD, YYYYNNNNMMMM  DD, YYYYSystemNC' MMM D, YYMMM D, YY def/SystemNC. [HH]:MM:SS.00 [HH ]:MM:SS .00NC3MM/DD/YYYY HH:MM:SS MM/DD/YYYY HH :MM:SS  NCK MMM D, YYYYMMM D, YYYY def/SystemNCL MMMM D, YYYYMMMM  D, YYYY def/SystemNCM NN, MMM D, YYNN, MMM D, YY def/SystemNCNNN, MMMM D, YYYYNN, MMMM  D, YYYY def/SystemNCONNNNMMMM D, YYYYNNNNMMMM  D, YYYY def/SystemNCP D. MMM. YYYYD. MMM. YYYYDIN 5008 (EN 28601)NCQ D. MMMM YYYYD. MMMM  YYYYDIN 5008 (EN 28601)NCRMM-DDMM-DDDIN 5008 (EN 28601)NCSYY-MM-DDYY-MM-DDDIN 5008 (EN 28601)NCT YYYY-MM-DDYYYY-MM-DDDIN 5008 (EN 28601)NCUWWWWNCStandardStandarddNC#.##0,00#.##0,00SystemNC #.##0,00 CCC#.##0,00 CCCNC#.##0,-- DM;[ROT]-#.##0,-- DM #.##0,-- DM-#.##0,-- DMROTNC TT.MM.JJJJTT.MM.JJJJ def/SystemNCTT.MM.JJTT.MM.JJ def/SystemNCNNNNT. MMMM JJJJNNNNT. MMMM  JJJJSystemNC T. MMM JJT. MMM JJ def/SystemNC [HH]:MM:SS,00 [HH ]:MM:SS ,00NCTT.MM.JJJJ HH:MM:SS TT.MM.JJJJ HH :MM:SS  NC T. MMM JJJJT. MMM JJJJ def/SystemNC T. MMMM JJJJT. MMMM  JJJJ def/SystemNC NN, T. MMM JJNN, T. MMM JJ def/SystemNCNN, T. MMMM JJJJNN, T. MMMM  JJJJ def/SystemNCNNNNT. MMMM JJJJNNNNT. MMMM  JJJJ def/SystemNC T. MMM. JJJJT. MMM. JJJJDIN 5008 (EN 28601)NC T. MMMM JJJJT. MMMM  JJJJDIN 5008 (EN 28601)NCMM-TTMM-TTDIN 5008 (EN 28601)NCJJ-MM-TTJJ-MM-TTDIN 5008 (EN 28601)NC JJJJ-MM-TTJJJJ-MM-TTDIN 5008 (EN 28601)NCWWWWNC' GeneralGeneraldNC' #,###.00#,###.00SystemNC('  #,##0.00 CCC#,##0.00 CCCNC)' $#,##0.--;[RED]-$#,##0.-- $#,##0.---$#,##0.--REDNC4'  MM/DD/YYYYMM/DD/YYYY def/SystemNC5' MM/DD/YYMM/DD/YY def/SystemNC6' NNNNMMMM DD, YYYYNNNNMMMM  DD, YYYYSystemNC7'  MMM D, YYMMM D, YY def/SystemNC>'  [HH]:MM:SS.00 [HH ]:MM:SS .00NCC' MM/DD/YYYY HH:MM:SS MM/DD/YYYY HH :MM:SS  NC['  MMM D, YYYYMMM D, YYYY def/SystemNC\'  MMMM D, YYYYMMMM  D, YYYY def/SystemNC]'  NN, MMM D, YYNN, MMM D, YY def/SystemNC^' NN, MMMM D, YYYYNN, MMMM  D, YYYY def/SystemNC_' NNNNMMMM D, YYYYNNNNMMMM  D, YYYY def/SystemNC`'  D. MMM. YYYYD. MMM. YYYYDIN 5008 (EN 28601)NCa'  D. MMMM YYYYD. MMMM  YYYYDIN 5008 (EN 28601)NCb' MM-DDMM-DDDIN 5008 (EN 28601)NCc' YY-MM-DDYY-MM-DDDIN 5008 (EN 28601)NCd'  YYYY-MM-DDYYYY-MM-DDDIN 5008 (EN 28601)NCe' WWWWNCBoeqeoePp, ~~2$:: SAP.AAPdddAPrddAPr HeaderS}AP.@APdddAPd dA<PN3T*Information Modelling using RDFA Pr FooterSAP.@APdddAP ddAJPNAT8( SA @A0APdE#S\AP.AAPdddAPrddAPdE#p ~2$:: S AP.AAPdddAPnddA PA Pr FooterSAP.@APdddAP ddAJPNAT8( SA @A0APdE#S\AP.AAPdddAPnddAPdE#Z~PSW5HDR.0AU1F C(569a(Build:5169)(SV569)]D!Address Book Fileaddress!9 Frameformat ZeichenformatTextformatvorlageStandard Text bodyText body indent Signature Salutation Marginalia List Indent Heading Heading 9 Heading 8 Heading 7 Heading 6 Heading 5  Heading 4  Heading 3  Heading 2 Heading 10  Heading 1Hanging indentFirst line indent(Title(SubtitleAbstractHeading prefaceHeaderAuthor detailsKeywords Contents Heading Index Contents 1 Contents 2 Contents 3 Contents 4 Contents 5 AddresseeCaptionEndnoteFooter Footer left Footer right Footnote Header left Header right0Horizontal Line0 List Contents0 List Heading0Preformatted Text0 QuotationsSender Frame contentsListTable Contents Table HeadingNumbering 5 StartNumbering 5 EndNumbering 5 Cont. Numbering 5 Numbering 4 StartNumbering 4 EndNumbering 4 Cont. Numbering 4 Numbering 3 Start Numbering 3 End Numbering 3 Cont.  Numbering 3Numbering 2 StartNumbering 2 EndNumbering 2 Cont. Numbering 2Numbering 1 StartNumbering 1 EndNumbering 1 Cont. Numbering 1% List 5 Start' List 5 End( List 5 Cont.&List 5! List 4 Start# List 4 End$ List 4 Cont."List 4 List 3 Start List 3 End  List 3 Cont.List 3 List 2 Start List 2 End List 2 Cont.List 2 List 1 Start List 1 End List 1 Cont.List 1 SuperscriptPreformatted text continued ReferenceText body blank beforePreformatted text hangingText body.hanging-indentNote Table Object index heading Object index 1Normal heading 3 Normal (Web)Numbering SymbolsBullet Symbols4Strong Emphasis5 Source Text Superscript3Citation8Variable Internet link9 Definition2EmphasisDefault Paragraph FontVisited Internet Link Line numbering FormulaRoot 124 125 126Standard First Page  Illustration TextDrawing Contents 6 Contents 7 Contents 8 Contents 9 Contents 10dHo1 4 5*yjK standard.dic soffice.dicZsun.dic@ IgnoreAllListY .Y g.Y .Y .yxTable of ContentsC <#><#><#><#><#><#><#><#><#><#>  !"#$ ux Table of Contents1Table of Contents PP3D D D.D P3D D D D P3D D D D P3D D D D P3D D D D P3D D D D P3D D D D P3D D D D P3 D D D D P3 D D D D   !"#$}s4}S)A  PAPdq s4|S)A  PAPd6NTKInformation Modelling using RDFS!APd dA P~T@5Constructs for Modular Description of Complex SystemsT Graham KlyneT:/BGaltimore Technologies, Content Security GroupTZ<1220 Parkview, Arlington Business Park, Theale, RG7 4SA, UK.w(T]Graham.Klyne@Baltimore.comS)A Courier NewA  wTAbstractTThis paper describes some experimental work for modelling complex systems with RDF. Basic RDF represents information at a very fine level of granularity. The thrust of this work is to build higher-level constructs in RDF that allow complex systems to be modelled incrementally, without necessarily having full knowledge of the detailed ontological structure of the complete system description. TThe constructs used draw on two central ideas: statement sets as contexts (based in part on ideas of McCarthy and Guha[1,2]) to stand for a composition of individual RDF statements that can be used in certain circumstances as a statement, and a system of "proper naming" that allows entity prototypes to be described in a frame-like fashion, but over a wider scope than is afforded by class- and instance- based mechanisms.A8 w|qwfsTG<Keywords: RDF, metadata, information modelling, semantic webT T%Table of ContentsK T1 1. Introduction 2SA@%..T1!1.1 Terminology 3SA@%. TJ!*1.2 Descriptive conventions and notation 3SA@%. T> 2. Motivation and background 5SA@%..TJ!*2.1 Resource Description Framework (RDF) 5SA@%. T@! 2.2 Describing complex systems 6SA@%. TL!,2.3 Descriptions based on generic entities 7SA@%. TS!32.4 Relationship to other semantic web activities 9SA@%. T3 3. Statement sets 9SA@%..TO!/3.1 A general container for sets of resources 9SA@%. TD!$3.2 Containers for statement sets 10SA@%. T=!3.3 Statement set notation 10SA@%. T. 4. Contexts 11SA@%..TH!(4.1 Representation of contexts in RDF 12SA@%. TS"34.1.1 Asserting statements in different contexts 13SA@%. TJ!*4.2 Using contexts: logical constructs 14SA@%. T?"4.2.1 Contexts as containers 14SA@%. T?"4.2.2 Contexts as statements 15SA@%. T>"4.2.3 Contexts as resources 16SA@%. T6 5. Generic entities 18SA@%..T<!5.1 URIs and Proper Names 18SA@%. T:!5.2 Proper names in RDF 19SA@%. TQ!15.3 Using proper names in generic descriptions 20SA@%. TB "6. Conclusions and further work 21SA@%..T6 7. Acknowledgements 22SA@%..T9 8. References 22SA@%..K T*1. IntroductionSA PTThis paper suggests some ways of modelling complex systems with RDF[3]. Basic RDF represents information at a very fine level of granularity, and the RDF description of a system of significant complexity can be very difficult to design and construct. The thrust of this work is to build higher-level constructs in RDF that allow complex systems to be modelled incrementally, without necessarily having full knowledge of the detailed ontological structure of the complete system description.A8CF:eTThese ideas are being used in experimental projects to describe and analyse systems consisting of complex interacting components, such as networked computer systems and software packages[19]. One could imagine these techniques being used to describe complex machines like cars, or social processes and interactions. We also plan to use some of these ideas to perform trust modelling.A8 qTS.The constructs used draw on two central ideas:SA PA PT4statement sets as contexts, based in part on ideas of McCarthy[1] and Guha[2], to stand for compositions of individual RDF statements that can be used or referenced as a single entities, andS2APdddA @ 373160971A8 >AqA8 JMq3w6FT5a system of proper naming, loosely derived from some ideas by Pat Hayes[13], that allows an entity to be referenced using a locally assigned identifier rather than a global URI. This is used to create prototype descriptions in which the assigned name can reference a characteristic of more than one entity.SRAPdddA PA @A @A @ 373160971A8 IMq3TzKThe goal is to use these constructs to introduce modularity into complex system descriptions. One might view RDF as an assembly language for metadata representation; this work builds higher level descriptive constructs on the RDF base, analogous to the use of functions and subroutines in mathematics and programming languages.S$A PA @A @TThis work overlaps semantic web ontological developments such as OIL and DAMLO, but has an emphasis on more easily assembled constructs for expressivity in RDF rather than semantic web inference and proof.TtZComments are welcome, and may be directed to the RDF-interest email distribution list[14].A8 UYqT1.1 TerminologyTpVStatement set: a collection of reified statements, whose container is an RDF resource.A8 uTContext: a statement set, with additional properties capturing certain structural and logical properties associated with contexts, described in this document.A8 uA8 rTLifting rule: any rule that allows us to deduce that truth of some statement in a context from the truth of another statement in a related context.A8 uT' Ontology is unhelpfully defined by my dictionary as the branch of metaphysics dealing with the nature of being. As a computer systems engineer I find it more helpful to think of ontology as the application of data typing mechanisms to arbitrary objects and concepts.A8 uT}6Ontological structure is the set of classes or types used in a description of some concept, and the relationship between them. RDF uses classes to indicate categories of objects of concepts. Classes may be related through multiple inheritance, and an RDF resource may be an instance of any number of classes.A8 uA8 uA8 rA8 uT3(1.2 Descriptive conventions and notationT%This paper is about structures in RDF. It focuses entirely on the graph syntax of RDF, described in section 5 of the RDF Model and Syntax specification[3], and does not make any assumption about the particular RDF serialization format used.SA PA @A8 qTThis note was originally drafted before N3 (Notation 3) [25] was widely known. Future versions may use N3 in place of the notation described here.SA @A8 8<qT&In describing RDF statements, square brackets are used to denote RDF resources, containing a name for the resource, or quoted strings to describe literal object values. RDF properties are shown as labelled arrows from subject to object:S.A PA PA @A @T71,[SubjectName] --propertyName--> [ObjectName]T aT;a0[SubjectName] --propertyName--> Literal-stringTj_RDF classes are given names starting with an upper case letter, RDF properties with lower case.TSpecific resource identifiers may be replaced by the expected rdf:Type of the resource in angle brackets. Thus, a general RDF statement might be described as:A8 >FpA8 >ATl17[] ----> []A8 A8 A8 (,T\QTo give a name to the reification of a statement, the following notation is used:TS1:Statement-id: [SubjectName] --propertyName--> [ObjectName]A8  Twhich can be interpreted as introducing a resource identified as Statement-id, and the following additional triples, corresponding to a reification of the statement:A8 AMpTf11[Statement-id] --rdf:type-------> [rdf:Statement]A8  A8 A8 #&TWa0[Statement-id] --rdf:predicate--> [propertyName]A8  A8 TVa/[Statement-id] --rdf:subject----> [SubjectName]A8  A8 TUa.[Statement-id] --rdf:object-----> [ObjectName]A8  A8 TxmTo apply a property to the reification of each of a set of statements, the statements are enclosed in braces:T#1[Resource] --property-->Ta {T(a Statement-setA8 Ta }TvThis shows that the reification of each member of Statement-set appears as the object of an RDF statement of the form:A8 2?pA8 2?A8 <?pT=1$[Resource] --property--> [Statement]A8 #TThe namespace prefix rdfc: is used for RDF classes and properties introduced by this paper to represent container and context information. The namespace prefix rdfn: is used for RDF elements introduced to represent naming information. Prefixes rdf: and rdfs: are used for namespaces defined in the RDF Model and Syntax[3] and RDF Schema Specification[4]. RDF resources and properties without prefixes are used in hyothetical examples with no defined relationship to any specific namespace.SA PA8 pA8A8 A8A8 A8 pA8A8 A8A8 A8 pA8 A8 pA8 A8 DGqA8 dgqT'2. Motivation and backgroundTSHThe main goals addressed by this work are to define RDF mechanisms that:T]create information models using higher level elements composed from basic RDF statements, andS2APdddA @ 2948564813Txdescribe frame-like generic models that can be used as prototypes to simplify the description of more specific entities.S2APdddA @ 2948564813TThese are required to be usable in an incremental, top-down fashion, where higher level descriptions and usage do not need to be aware of the details of lower level ontological structures.T3(2.1 Resource Description Framework (RDF)TResource Description Framework (RDF)[3] is a W3C recommendation for a standard representation of metadata, based on ideas with roots in knowledge representation research conducted over the past 30 years or so. The specification defines an abstract directed labelled graph model for RDF, and an XML-based serialization (there is some discussion about defining other serialization formats for various purposes). The nodes of this graph are RDF resources, and the arcs are RDF properties. A companion RDF Schema specification[4] describes how RDF may itself be used to define a type system based on RDF classes, and constraints on the ways in which RDF classes and properties may be combined in a description.A8 $'qA8 qTwThe importance of RDF is not that it is demonstrably better than any other form of knowledge representation, but that it has a reasonable chance of becoming a widely used Internet standard, and that it is designed for use in an open web environment. To exchange information (as opposed to raw data) between computer systems or applications requires agreement about its representation. To this end, I build upon, rather than extend or modify, the core concepts of RDF, and the resulting framework is fully expressible within the graph structure of basic RDF. This adherence to basic RDF structure does not constrain the internal wortkings of implementations; for example, reified RDF statements may be represented in ways that are far more efficient than that suggested by basic RDF. The relationship to basic RDF structure is needed when exchanging information between applications.TSThe RDF specification is felt by some to be confusing as an ab initio guide to RDF, and there are several papers on the web that provide a gentler introduction. One such introduction that may be particularly helpful is by Pierre-Antoine Champin[23], which starts by explaining the RDF graph model that is the underlying basis for the constructs described here. A topic that often causes particular confusion about RDF, and which is quite fundamental to this paper, is reification: this is the construct that allows us to make statements about statements, by constructing an RDF resource to stand for any given RDF statement. Thus, we can have a resource identifier that stands for an RDF statement, and using that identifier we can make statements about that statement.S$A PA @A @A8 <EA8 qT)2.2 Describing complex systemsTpZThe structures described here evolved from an attempt to describe complex system components in RDF, from which it was required to draw inferences about the compatibility of various combinations of component. Originally, we tried to construct meaningful descriptions using the basic RDF property-subject-object constructs, but found it was very difficult to predict the required graph structure without first assembling all of the information to be described. As new factors came into play, we were finding that existing structures had to be redesigned. Later, we approached the same problem using an informal high-level abstraction to stand for a component description consisting of arbitrary RDF statements. In this way, we found that high-level interactions between components could be captured, and subsequently refined with more detailed information.w TkSuppose I wish to describe the kind of fuel needed to operate my car. A statement like this might be used:SA PA PT)1[MyCar] --fuelType--> DieselTmeaning that my car runs on diesel fuel. This is just fine for determining which pump I need to stop at when refuelling. But for an engineer dealing with the car as a piece of machinery, more detailed information is needed, and the fuelType might be considered a property of of the cars engine, rather than of the car as a whole. The engine might be replaced with one using a different kind of fuel. At this level of detail, the model suggested above is not adequate, and a different structure is needed; e.g.A8 pTF1;[MyCar] --hasEngine--> [MyCarEngine] --fuelType--> DieselTV)This simple example shows how simple RDF model ontological structures may need to be adjusted to take account of new levels of detail. If, however, we treat the description of my car as a collection of statements, then their relationship to the overall entity that is my car may be more flexible:S"A PA @A @TE1[MyCar] --description-->S"A PA @A @T<a {S.A PA PA @A @T0a :S"A PA @A @TPa# [FuelUsed] --fuelType--> DieselS"A PA @A @T0a :S"A PA @A @Ta }SA PTFOn more detailed examination, this might turn out to be structured as:S.A PA PA @A @T11[MyCar] --description-->SA PTa {SA PT<a :S.A PA PA @A @T4a [Engine] --description-->SA PTa {SA PTa :SA PT>a% [FuelUsed] --fuelType--> DieselSA PTa :SA PTa }SA PTa :SA PTa }SA PT5*2.3 Descriptions based on generic entitiesTAnother recurring idea is that entities are often variations on a generic theme. A car can be broadly classified in terms of a make and model, with variations of body colour, engine type, equipment levels and optional accessories. On a manufacturing level, what is outwardly the same vehicle type may be produced in a number of subtlely different forms as product improvements or economies are devised.T~To effectively manage descriptions of such generic components, it is important to be able to refer to a common description, and then describe local variations from the base description. This idea is fundamental to CC/PP[5], one of the first protocols to be based on RDF, which uses standard descriptions and local variations to describe web client capabilities and characteristics.A8 qTfIn classic AI and knowledge representation research, one way this effect is achieved uses frames. Minsky[6] describes a frame as a remembered framework to be adapted to fit reality by changing details as necessary. More recently, building on the basic framework of RDF schemas[4], ontological work with RDF[8,9] has implemented similar structures by associating slots and slot constraints with RDF classes. Thus, an instance of an RDF class is presumed to have an associated default value for each of the defined slots. Where a specific value is not specified for a given instance, a value may be supplied by the slot description in the class definition. CC/PP[5] adopts an ad-hoc approach using a specific RDF property (default) that exploits the regular structure of CC/PP profiles.A8 knqA8 qA8 8=qA8 qT"A drawback with this approach is that it presumes a well defined relationship between the frame and the slots. If it is not known that an given property is attached to a specific type of resource (c.f. The fuelType example from the previous section), then there is no clear rule to state where the default value may be found. If the ontological structure is revised, the location of the slot value may change, which works against the goal of incremental modelling without full knowledge of the ontologies involved.A8 pTThe framework described later allows more flexible linkage between a slot value and a prototype. Rather than depending on a fixed ontological relationship between a slot value and its corresponding definition, arbitrary linkage is allowed through the use of a framework of proper names, based loosely on some ideas put forward by Pat Hayes[13]. These names are used to link slot definitions to slot values.S$A PA @A @A8 VZqT8-Additional structures are used here because RDF has no concept of a local name: all properties and resources are names using URIs, which are, by design, a global namespace. There is no mechanism for a URI to denote one thing in one part of an RDF model, and to denote soemthing else in another part.TReturning to the example of [MyCar]: Many of the details about my car may be deduced with a reasonable degree of reliability by knowing the make and model:A8 #pT(1[MyCar] --isa--> [FordEscort]T`FA description of [FordEscort] might include the following information:A8 pT(1[FordEscort] --description-->T(a {SA PA PT<a :S.A PA PA @A @T4a [Engine] --description-->SA PTa {SA PTa :SA PT>a% [FuelUsed] --fuelType--> PetrolSA PTa :SA PTa }SA PTa :SA PTa }SA PTFrom the above statements, it might be inferred that my car runs on petrol. But what if my car is atypical, and has a diesel engine?T4According to RDF, [FuelUsed] is a specific resource with a given, globally unique URI. There is no way to allow the resource name [FuelUsed] to mean different things in the context of different instances of [FordEscort]. Therefore, I cannot just say:A8 pA8 pA8 pT61[MyCar] --isa--> [FordEscort]SA PT1a[ ] --description-->SA PTa {SA PT<a :S.A PA PA @A @T4a [Engine] --description-->SA PTa {SA PTa :SA PT>a% [FuelUsed] --fuelType--> DieselSA PTa :SA PTa }SA PTa :SA PT(a }SA PA PTu because, using RDF, the resource named [FuelUsed] here is necessarily the same as the resource named [FuelUsed] in the description of [FordEscort], it cannot be used to indicate that the fuel type for my car is different from the normal fuel type for a [FordEscort].S$A PA @A @A8 '1pA8 eopA8 pA8 pT<12.4 Relationship to other semantic web activitiesTQThis work overlaps semantic web ontological developments such as OIL[8] and DAMLO[9], in that it defines RDF structures with inference properties, but has an emphasis on defining constructs for expressivity in RDF rather than semantic web inference and proofs. The proposals here do not depend on a full framework of first order logic.A8 DGqA 0P-A8 RUqTD9It is not a goal of this work to provide logical capabilities beyond those provided by the ontological work: any information using the structures described here should be expressible using the ontological frameworks being developed by other groups. In due course, mechanical mapping processes should be devised.TThat it is possible to write any program in low-level machine code doesnt make it a desirable or useful way to construct large programs. Similarly, this work aims to provide higher level constructs than basic RDF statements for assembling RDF models of complex systems.TQ(As care has been taken to ensure that the entire framework is grounded in the standard RDF[3] and RDF schema[4] specifications, it should be usable with any of the generic RDF handling tools being developed. It is agnostic with respect to API, operating environment and RDF serialization format.A8 Z]qA8 loqT3. Statement setsTshRDF defines a way to represent collections of statements which suffers from some practical difficulties:TThe rdf:Bag container class and associated containment properties make it diffcult to add new statements to a collection without knowing all of the statements already belonging to that collection.S2APdddA @ 373529031A8  p3TEIt is not possible to use subproperties to represent different containment relations for a single container, because each member has a different containment relation property (rdf:_1, rdf:_2, etc.).S2APdddA @ 373529031A8 pA8 A8 pA8 3T;The standard container classes have no way to represent distributive referents within an RDF graph. The construct rdf:aboutEach is an XML serialization form that must be expanded in the corresponding RDF graph model.S2APdddA @ 373529031A8 spA8 sv3T8-3.1 A general container for sets of resourcesTgA container class and property are defined, overcoming the above problems when used for statement sets:SA PA PTNrdfc:Set is a class that may be a container for an arbitrary set of resources.SRAPdddA PA PA PA @ 373504971A8 pA8 3TBrdfc:member is a property whose domain is an rdfc:Set, and whose range is any RDF resource type. It is used to indicate that its object is a member of its subject container.SHAPdddA PA @A @ 373504971A8 pA8 A8 -5pA8 -13TThus, we have:TN1'[Container] -rdf:type-----> [rdfc:Set]A8 A8 "TDa+[ ] --rdfc:member--> [SomeResource]A8 TGa.[ ] --rdfc:member--> [AnotherResource]A8 Ta[ ] :T.a etc.A8 TNote that, unlike the standard RDF container classes, this structure does not permit repeated instances of the same resource or value in an rdfc:Set container.A8 pA8 T-To maximize compatibility with RDF schema[4]:SHAPdddA PA PA @ 373428231A8 ),q3T;rdfc:Set is defined to be a subclass of rdfs:Container, andS<APdddA PA @ 373428231A8 pA8 A8 (6pA8 (,3TMrdfc:member is defined to be an instance of rdfs:ContainerMembershipProperty.S=APdddA @A @ 373428231A8 pA8 A8 ,LpA8 ,03T,!3.2 Containers for statement setsT|qBuilding on the generic set container described above, a statement set and membership thereof are represented by:T[rdfc:StatementSet is a subclass of rdfc:Set, and is a collection of reified RDF statements.S2APdddA @ 373445421A8 pA8 A8 #+pA8 #'3Trdfc:quotes is defined to be a sub-property of rdfc:member, used to indicate a reified statement that is a member of a statement set. Its domain is restricted to rdfc:StatementSet, and its range is rdf:Statement.S2APdddA @ 373445421A8 pA8 A8 /:pA8 /3A8 pA8 A8 pA8 3T3Thus, we have:SA PA PTP1)[SS] -rdf:type-----> [rdfc:StatementSet]A8  A8 T>a%[ ] --rdfc:quotes--> [SomeStatement]A8  TAa([ ] --rdfc:quotes--> [AnotherStatement]A8  Ta [ ] :T'a etc.A8  T%3.3 Statement set notationTThe brace notation, used in this paper for the purposes of exposition, avoids the need to invent an explicit reified statement identifier for describing statement sets, so we can write the following:T81[StatementSet] --rdfc:quotes-->A8 Ta {T%a [Sub1] --prop1--> [Obj1]T%a [Sub2] --prop2--> [Obj2]Ta :T!a (etc.)A8 Ta }TThe corresponding RDF must still use reified statement identifiers, but these are assumed to be automatically generated and not interesting for the purposes of constructing higher level information models.SA PA @T 4. ContextsT7The idea of contexts was proposed by McCarthy[1,10], and some more detailed theory and applications were developed by R. V. Guha in his PhD thesis[2]. A context is characterized by the fundamental relationship is true in, or ist, where:SA PA8 -3qA8 qA8 pTC1[Statement] --ist--> [Context]SA PA PTzCmeans that [Statement] is true in [Context]. A context may be taken to be an environment within which some statements are held to be true. Thus, to use a statement in some chain of reasoning, one must determine (or assume) that a context in which the statement is true in applies to the circumstances of that reasoning.SA PA8 pA8 "+pTGuha describes a context as defining its own language, in the sense that it establishes a framework of meaningful expressions based on first order logic. For the purposes of modelling with RDF, the first order logic is replaced by the structure of basic RDF assertions, possibly taken together with some specified RDF properties and associated set of rules that define a language and deductive framework for that context. Tim Berners-Lee describes in The Semantic Toolbox[12] one possible form of RDF properties and associated rules to describe first order logic. It seems useful to allow that the applicable RDF logic rules may be associated with a context.A8 A8 qTE:Thus, RDF provides the basic vocabulary and grammar for representing assertions associated with contexts, and RDF schema may be used to describe ways the vocabulary is used, but rules of logical deduction associated with a context (beyond the minimal rules defined by RDF) need to be established by external means.TAnother key feature of contexts is the idea of lifting rules or lifting axioms: rules to deduce the truth of statements in one context from the truth of statements in some other context. In the absence of applicable lifting axioms, such deductions are not typically valid. I envisage that the topic of lifting between RDF contexts can be approached gradually, through the introduction of specific types of relationship between contexts, rather than by adopting a fully generalized theory of lifting.T McCarthy describes a context as an abstract object. This suggests that one can make assertions about contexts as well as within contexts. He also describes a context as a generalization of a collection of assumptions. These assumptions may concern the vocabulary and language used as well as any facts assumed; not all assumptions may be explicitly known. McCarthy also points out that, in order to work within a framework of first order logic, statements must be reified to associate them with contexts. RDF provides a mechanism for reification of statements, but it tends to be cumbersome. For the purpose of discourse the needed reification of RDF statements is implicit in the notation used.S"A PA @A @A8 bgvA8 |vT0%4.1 Representation of contexts in RDFTb?A context represented in RDF has the following characteristics:SA @A @T)It is a collection of reified statements.SFAPdddA @A @A @ 3733405013T<The collection and the reified statements are RDF resources.SFAPdddA @A @A @ 3733405013T\The statement resources are explicitly related to the collection resource by RDF properties.SFAPdddA @A @A @ 3733405013TgDifferent properties may be used to indicate different relationships between collection and statements.SFAPdddA @A @A @ 3733405013TVContexts may have properties that tell us something about the statements they contain.SFAPdddA @A @A @ 3733405013T:Contexts may be related to other contexts in various ways.SFAPdddA @A @A @ 3733405013TThese characteristics are expressed by describing a Context as a statement set with some additional structural and logical properties:SHAPdddA PA PA @ 373340501A8 4;rA8 ANr3Trdfc:Context is a subclass of rdfc:StatementSet, and represents a context. By inheritance this consists of a set of reified statements.S2APdddA @ 373340501A8 pA8 A8 /pA8 "A8 #/p3TBrdfc:asserts is a sub-property of rdfc:quotes, indicating a reified statement that is a member of a context, and which is also asserted to be true in that context. Thus, it corresponds to the ist (i.e. is true in) relation described by McCarthy and Guha. It has a domain of rdfc:Context, and a range of rdf:Statement.SHAPdddA PA @A @ 373340501A8 pA8 A8 "-pA8 "&A8 #pA8 A8 4ApA8 473 T3Thus, we have:SA PA PTU1.[SomeContext] -rdf:type------> [rdfc:Context]A8 A8 !%THa/[ ] --rdfc:asserts--> [SomeStatement]A8 TKa2[ ] --rdfc:asserts--> [AnotherStatement]A8 T a[ ] :T0a etc.A8 TXThe rdfc:quotes property may be used with an rdfc:Context to mean that the indicated statement is included in the context, but is not asserted by the context to be true. This is needed if the context makes statements about other statements that are not themselves held to be true; e.g. Graham says the sheep is pink might be represented as:S.A PA PA @A @A8 pA8 A8 -9pA8 -1Tk1*[Context] -rdf:type------> [rdfc:Context]SA PA PA8  A8 !T"a [ ]SA PTBa[ ] --rdfc:asserts-->SA PA8  T(a[ ] {SA PTGa.[ ] [Graham] --says--> [SheepIsPink]SA PT(a[ ] }SA PTa [ ]T3a[ ] --rdfc:quotes-->A8  Ta[ ] {TDa9[ ] SheepIsPink: [The sheep] --colour--> pinkTa[ ] }TfeNOTE: it might be questioned whether non-asserted statements actually belong in a context. One of the motivations for this work is to have a way to collect together related parts of an RDF graph, so this has been adopted as a helpful idea. Also, given the way that contexts are constructed from statement sets, it is a natural and easy structure to adopt.A8A8 T;04.1.1 Asserting statements in different contextsTStatements may be associated with multiple contexts, with their various assertions conveying different kinds of information. Suppose that I make the following statements:SA PA PT.1#[MyCar] --fuelEconomy-----> 40mpgT/a$[MyCar] --engineCapacity--> 1600ccT"The first statement may be something that I believe. The second statement may also be something that I must assure for the purposes of obtaining insurance. Contexts corresponding to statements I believe and statements I assure can be constructed thus:SA PA PTD1[MyBeliefs] --rdfc:asserts-->SA PA8 Ta {SA PT@a' [MyCar] --fuelEconomy-----> 40mpgSA PTAa( [MyCar] --engineCapacity--> 1600ccSA PTa :SA PT*a }SA PA PT(andSA PA PTG1 [MyAssurances] --rdfc:asserts-->SA PA8 Ta {SA PTAa( [MyCar] --engineCapacity--> 1600ccSA PTa :SA PT*a }SA PA PTLNote that a statement may be true in any number of different contexts. Thus, the rdfc:asserts property may be applied any number of times using a given statement. (Naturally, any number of statements may be asserted by a given context: this grouping of statements into larger units is one of the motivations for these proposals.)SA PA PA8 R^pA8 RVT2'4.2 Using contexts: logical constructsT'4.2.1 Contexts as containersTAn obvious use of a context is as a container for some collection of statements. An RDF document may contain a number of statements. Assertions applied to the document might be taken to apply to each of the statements contained in the document.TFor example, a signature applied to a document creates some kind of assurance about the content of that document. This assurance would reasonably be considered to apply to each of the statements within the document. To apply such assurances individually to each statement seems cumbersome, especially given the fine-grained nature of individual RDF statements. Thus, we have one application for contexts that is to facilitate making assertions about collections of statements.S$A PA @A @T].A simple approach that has been suggested would be to allow statements about a collection to be applied automatically to its members. This allows no way to distinguish statements about the collection itself from statements about its members. Suppose I use a context [MyWebSite] to contain a set of statements about my web site, and attach a property that says that certain people are authorized to make changes. Does this property apply to the collection of statements about the web site, or does it apply individually to each of the contained statements?SAP d dA8 pTHI propose a level of indirection that makes explicit the distribution of assertions over contained statements. It follows the idea of interpretation properties described by Tim Berners-Lee[12]. In this case, the proposed interpretation property is one that explicitly applies assertions to all of the contained statements:S+AP d dA PA PA8 qTJ1#[CarInsurance] --rdfc:asserts----->SA PA8 T-a[ ] {SA PTOa6[ ] [Car] --registration----> ABC123XSA PTNa5[ ] [Car] --engineCapacity--> 1600ccSA PT.a[ ] :SA PT-a[ ] }SA PTRa+[ ] --rdfc:applyToAll--> [_stmt]SA PA8 TaSA PT;a"[_stmt] --assuredBy--> [Car owner]SA PTBa)[_stmt] --assuredBy--> [Car registration]SA PTyHere, rdfc:applyToAll indicates quantification over statements within the context, and [_stmt] is a place-holder for such statements. For each statement contained within the context, and for each statement involving the place-holder resource [_stmt], a corresponding statement is asserted. In the above example, the statements of assurance by both the car owner and the car registration document are applied to the statements about registration mark and engine capacity that are contained within the context:S5AP d dA PA @A @A8 pA8  A8 W^pA8 pTLa'S1: [Car] --registration----> ABC123XSA PA PT?a&S2: [Car] --engineCapacity--> 1600ccSA PTa :SA PTaSA PT<a# [S1] --assuredBy--> [Car owner]SA PTCa* [S1] --assuredBy--> [Car registration]SA PT<a# [S2] --assuredBy--> [Car owner]SA PTCa* [S2] --assuredBy--> [Car registration]SA PTa :TThus, rdfc:applyToAll indicates that all properties applied to the object resource are to be applied individually to each statement that holds in the subject context. The resource [_stmt] is a local place-holder of implied type rdf:Statement, identified by an arbitrary URI-reference.A8 pA8 A8  A8 pA8 pA8 TfNOTE: rdfc:applyToAll is presented here as a one-off case for use with contexts. The idea can be generalized to apply to an arbitrary container class and member type[20].A8A8 A8 pA8  A8 qTfyNOTE: This is an alternative way to model statements about sets of statements than that described in the RDF Model and Syntax[3], section 4.2. A significant difference is that this approach represents the distributed referent structure within the RDF graph, and therefore can be applied to contextual information possibly not otherwise available when the RDF is de-serialized.A8A8 A8 }qT'4.2.2 Contexts as statementsTA motivation for this work has been to find ways of combining basic RDF statements into higher-level statements that can be treated as basic statements. A context can be treated as such a composite statement.SA PA @Tg>The reification of a statement is a structure that stands irrespective of whether the statement is true, and a statement set is a collection of reified statements that does not consider the truth of them. A reified statement is asserted in a context if it is considered to be true in that context. Similarly a context contains statements that are true under some circumstance, irrespective of whether that cirumstance applies. A context may be asserted in some other context, meaning that statements true that context are also regarded as true in the surrounding context.A8 rA8 rTF;In this respect, a context may assume some of the attributes of a statement: it may be regarded as true or false (or unknown) in the sense that the statements that it embodies are all true, or otherwise. Its truth (or otherwise) may depend upon the truth of some other context. The kinds of metalogical operations that can be applied to statements (conjunction, implication, negation, etc.) could be usefully applied to contexts, treating a context as a kind of composite statement. It can even make sense to combine statements and contexts in metalogical expressions.T These ideas for interchangeability of contexts and RDF statements can be expressed by creating a unifying superclass for them. Defining one as a subclass of the other is not satisfactory because each has properties not shared by the other. (A statement has a very specific structure and relationship to the RDF model; a context can be used in ways not applicable to a statement.) The class rdfc:Assertable is defined to be a superclass of both rdf:Statement and rdfc:Context, e.g. with RDF scheme statements of the form:S$A PA @A @A8 pA8 A8 pA8 A8 pA8 T[14[rdfc:Assertable] ---rdf:type---------> [rdfs:Class]A8 A8 Tna9[rdf:Statement] -----rdfs:subClassOf--> [rdfc:Assertable]A8 A8 A8 )-Tna9[rdf:StatementSet] --rdfs:subClassOf--> [rdfc:Assertable]A8 A8 A8 )-Tna9[rdfc:Context] ------rdfs:subClassOf--> [rdfc:Assertable]A8 A8 A8 )-TAWhere a statement, statement set or a context may be used as the domain or range of some property, this can be expressed in a schema by the rdfc:Assertable class. Thus, the ranges of rdfc:quotes and rdfc:asserts are revised, relaxing the initially specified object type of rdfc:Statement to allow statements of the form:A8 !#vA8 pA8 A8 pA8 A8 pA8 A8 pA8  pA8 Tk16[] --rdfc:quotes---> []A8 A8 A8 %)Tka6[] --rdfc:asserts--> []A8 A8 A8 %)TkcCwhere rdfc:Assertable can be a statement, statement set or context.A8 pA8  T&4.2.3 Contexts as resourcesT:/One of the difficulties I have observed in trying to use RDF to model real world situations is that it is very difficult to construct a model of meaningful complexity without becoming overwhelmed by the details of the RDF graph. The idea of contexts as containers leads quite naturally to a treatment of contexts as first class resources in their own right, representing the collection of statements describing some physical-world object. Further, the use of contexts as statements allows the statements contained within a context themselves to be contexts.T-"Using these ideas, and the extra RDF properties described above, one can construct models in terms of high-level resource and property concepts. The validity or consequences of these high-level relationships is discovered by looking inside the corresponding contexts to find the component parts and the local interactions between them. Thus, one of the goals of this work is realized, allowing high-level models of complex systems to be constructed without knowledge of the more detailed structures involved. Consider the description of a car:TE1 [MyCar] --manufacturer--> [Ford]SA PA PT;a"[ ] --model---------> [Escort]SA PT@a[ ] --rdfc:asserts-->SA PA8  T&a {SA PT>a% [Engine] --model--> [CVH]SA PTMa& [ ] --rdfc:asserts-->SA PA8 T3a {SA PTOa6 [Fuel] --fuelType--> petrolSA PTOa6 [Spec] --capacity--> 1600ccSA PT5a :SA PT?a }SA PA PTNa) [Body] --style--> HatchbackSA PA PTKa$ [ ] --rdfc:asserts-->SA PA8 T1a {SA PTMa4 [Shell] --material--> steelSA PTMa4 [ ] --weight----> 200KgSA PTIa0 [Doors] --count-----> 5SA PT2a :SA PT1a }SA PT'a :SA PT&a }SA PTThis example illustrates how a description of a complex object or system can be built up component-wise from contexts describing the various components. At each level, a context is used as a resource that stands for the entity being described. Statements that are clearly about the resource as a whole can be attached to the resource directly (e.g. manufacturer above), while statements about properties of component parts can be associated with the appropriate part in a way that is not invalidated if the internal structure needs to change. For example, consider fuelType, which is associated above with the engine type. Suppose the engine description is restructured to consist of mechanical, fuel and electrical subsystems:A8 _kpA8 8@pT>1[Engine] --model--> [CVH]SA PA PTAa[ ] --rdfc:asserts-->SA PA8  T'a {SA PTGa. [MechSubSystem] --rdfc:asserts-->SA PT*a {SA PTFa- [Cylinders] --count-----> 4SA PTJa1 [ ] --capacity--> 400ccSA PTFa- [Valves] -----count-----> 8SA PTMa4 [CamShaft] ---location--> OverheadSA PTQa8 [ ] ---drive-----> Toothed beltSA PT+a :SA PT*a }SA PT9a. [FuelSubSystem] --rdfc:asserts-->Ta {T:a/ [Fuel] --fuelType----> petrolT=a2 [ ] --systemType--> injectionTa :Ta }T9a. [ElecSubSystem] --rdfc:asserts-->Ta {TBa7 [Ignition] --systemType--> electronicTa :Ta }Ta :Ta }TFFor the purposes of refuelling, the assertion [Fuel] --fuelType--> petrol is sufficient. This statement can be inferred in exactly that form, even though the internal structure of the description of the car has been changed. I believe that this ability to describe and infer certain facts independently of the detailed ontological structure is crucially important to the practical construction of models describing complex systems. Some initial exercises have suggested that using contexts in this way very greatly improves our ability to design models of real world objects.S$A PA @A @A8 /LpT5. Generic entitiesT|SSection 2.3 has introduced the motivation for generic entities, and the difficulties of using ontologically based frame descriptions. Here, I describe a system of proper naming for RDF that is loosely based on some ideas put forward by Pat Hayes[13] on the RDF logic mailing list[14], and illustrate its use to construct generic models.A8 qA8 qT$5.1 URIs and Proper NamesTRDF uses URIs and URI-references, defined by RFC 2396[16], to name resources. URIs are generally defined to be globally unique: a URI that appears in different places is always expected to indicate the same resource (which does not necessarily mean the same data, but that is not pertinent to the present discussion). URI references have a relative form, but RFC2396 is quite clear that this is a lexical convention and that the relative form should map to an absolute (globally unique) form depending on the lexical environment in which it occurs. Typically, the URI of a containing document is used to construct the global URI corresponding to a relative URI.A8 59qTPat Hayes observes[13] that much human discourse does not depend on global names to identify the topics being discussed; rather, proper names are used that have accepted meaning within some context of discourse. For example, in a discussion that alludes to American geography, Boston is typically understood to be a city in the US state of Massachusetts, but in other contexts it might mean something else. Here, the meaning of Boston is understood through a binding indicated by social context.A8 qTrIn logic, mathematics and computer programming, local names are often used to stand for a concept within some defined environment. The fact that the local name is not bound to some global meaning gives rise to its expressive power. Formal parameters of functions and procedures are responsible for much of the expressive power of notations used in logic, mathematics and computer programming. There has been discussion about how to introduce quantification into RDF to extend its expressive power (e.g. Tim Berners-Lees Semantic toolbox[11], and many others); there is no question that some such mechanism is required.A8 $qTUsing the social concept of proper names for local variables may seem strange; I suggest that this is an appropriate unifying of mechanisms, allowing a continuum of locally bound names (in expressing purely logical constructs) to socially bound names (in descriptions of ideas with social scope). Specifically, ideas that start out as being socially defined may be carried into contexts where the definition is logically bound. For example, the name Boston may be introduced in a social context with purely social binding, but may be carried over into a context of air transport where it is ultimately bound into a mathematical framework of spatial coordinates that ultimately define where an airplane must land.S$A PA @A @T"5.2 Proper names in RDFTt6This section proposes an experimental mechanism for expressing proper names within standard RDF. A proper name may be an arbitrary string of characters. International proper names, required to support social naming conventions, are accommodated by allowing characters from the UCS-4 (Unicode)[22] repertoire.S$A PA @A @A8 &*qTOne or more proper names may be associated with any RDF resource by the rdfn:properName property. The domain of rdfn:properName is any RDF resource, and its range is a literal string containing the proper name value:A8 HWpA8 HLA8 qpA8 quT>1%[MyCar] --rdfn:properName--> my carA8  T>a%[ ] --rdfn:properName--> ABC123XA8  T<a#[ ] --rdfn:properName--> EmilyA8  T-These three properName properties indicate different names by which [MyCar] may be referenced: in informal conversation, I talk about my car, in formal communications (e.g. insurance application) I may refer to it by a registration mark ABC123X, and my wife might refer to it in an anthropomorphic style as Emily. Clearly, someone else might use the name my car to refer to a completely different vehicle. In general, a proper name string is not bound to unique entity outside of a given context of usage.A8 pA8 DKpTA particular context in which a proper name applies may be indicated by including the rdfn:propername statement in that context.A8 VepA8 VZTfjNOTE: the resource [MyCar] above is an RDF resource that is generally assumed to have a globally unique identifier in the form of a URI-reference. It may be that the URI-reference is a generated value like a uuid: or cid: URI with no obvious relationship to any specific entity. There has also been some discussion in the W3C RDF interest group[15] about anonymous resources (without names, or having names that are explicitly not globally unique URIs). This is a matter for continuing debate. In any case, the rdfn:properName property associates the indicated name string with an RDF resource, however identified.S$A PA @A @A8A8 A8 pA8 pA8 pA8 Z^qA8 pA8 T4An inference rule associated with rdfn:properName tells us that two resources with the same proper name value are equivalent. That is, any property true of one such resource may be taken to be true of the other. Thus, proper names provide a logical framework for asserting the equivalence of two resources.S$A PA @A @A8 "1pA8 "&T[fNOTE: this description skirts an issue of the relationship between resources and resource identifiers. One school of thought asserts that the relationship between resources and identifiers is 1:1-onto; i.e. each resource identifier indicates exactly one resource, and each resource has exactly one resource identifier. Another view allows multiple resource identifiers to indicate the same resource. This debate is avoided by talking about equivalence of resources while remaining agnostic on the issue of sameness.A8A8 A8 rA8 rA8 rT9.5.3 Using proper names in generic descriptionsTSuppose my car is a Ford Escort. From this assertion, one can immediately infer a range of facts about my car as Ford Escort is an established kind of car. Proper names allow us to construct RDF descriptions that capture this kind of prototype-based inference.T71[FordEscort] --rdfc:asserts-->A8 Ta {TSa: [FordEscortBody] --rdfn:properName--> Ford Escort bodyA8 T=a$ [ ] --rdfc:asserts-->A8 Ta {T'a (details about car body)Ta }TWa> [FordEscortEngine] --rdfn:properName--> Ford Escort engineA8 T?a& [ ] --rdfc:asserts-->A8 Ta {T)a (details about car engine)Ta }Ta :Ta }TK1&[MyCar] --rdfc:asserts--> [FordEscort]SA PA PTa {SA PT\a5 [MyCarBody] --rdfn:properName--> Ford Escort bodySA PA8 T8a [ ] --rdfc:asserts-->SA PTa {SA PT:a! [BodyPaint] --colour--> RedSA PTa :SA PTa }SA PTa :SA PT(a }SA PA PTEThe above examples show a context [FordEscort] that describes a generic Ford Escort car, and another context that describes my car as a kind of Ford Escort. The generic Ford Escort is defined in terms of specific RDF resources ([FordEscortBody], [FordEscortEngine], etc.) that stand for components of the generic car type. Thus, of itself, the resource [FordEscortBody] does not stand for a specific car body, but expresses properties common to all Ford Escort car bodies.S$A PA @A @A8 ".pA8 pA8 pA8 cspTmThe description of [MyCar] asserts all of the statements that are true of a generic Ford Escort. It also asserts that [MyCarBody] has a proper name the same as [FordEscortBody]. The inference rules associated with rdfn:properName then allow us to deduce that, in the context of [MyCar], all statements about [FordEscortBody] are also applicable to [MyCarBody]. Additional statements about [MyCarBody], such as it having red paint, also apply in this context.A8 pA8 wpA8 pA8 pA8 A8 vA8 pA8 6FpA8 ^ipA8 pT*6. Conclusions and further workT~ORDF mechanisms based on statement sets and contexts have been described that allow descriptions of complex systems to be constructed without necessarily having detailed knowledge of the ontological structure of the system components used. I believe this is a key enabler for the practical construction of complex system models in RDF.S$A PA @A @T5*An additional RDF mechanism has been described, based on the idea of a proper name, that allows descriptions to be based on prototypes, in a fashion similar to the frame based descriptions proposed by AI researchers, but not depending on detailed knowledge of the ontological structures involved.T=Articulated visions for the Semantic Web require that anyone must be able to say anything about anything[21]. It is unreasonable to expect everyone to adopt exactly the same ontological structure for making statements about an entity; apart from political and perceptual differences, that approach cannot scale. This leads to my assertion that practical modelling of complex systems requires statements that can stand independently of finer ontological details. This is not a dismissal of ontological structures; work on onological frameworks such as OIL[8] and DAMLO[9] is needed to underpin verification of web-based information. In due course, I would expect a theory to emerge that relates descriptions based on incomplete ontologies to more rigorously complete frameworks. I view basic RDF as a kind of assembly language for information modelling, and see this use of contexts and proper naming as a parallel to procedures and formal parameters in programming languages, used to aid the construction of complex object descriptions without adding new formal capabilities.A8 hlqA8 /2qA 0;-A8 =@qTk`The constructs presented here are being used in the following ongoing experimental developments:T{'A graphical tool for RDF modelling[18].S2APdddA @ 373640271A8 "&q3T,An experimental RDF-driven expert shell[19].S2APdddA @ 373640271A8 '+q3TaWe also aim to develop mechanisms for trust modelling and inference; modelling social trust structures and overcoming the brittleness of purely cryptographically based approaches to trust in e-commerce, etc. Another area for investigation is the design of mechanisms for managing non monoticic reasoning, and other logical extensions of contexts [20].S$A PA @A @A8 \`qT+In messages to the RDF interest group, Dan Brickley has proposed[17] an alternative approach to labelling anonymous RDF resources; i.e. resources whose formal URI or URI reference is unknown. The outcome of these discussions may affect the exact form of naming preferred.A8 @DqT7. AcknowledgementsT8This note is based in part on an earlier document[20], for which I received valuable feedback from Brian McBride, Dan Brickley, Jan Grant and others from HP Labs and ILRT in Bristol, UK. The ideas have also benefited from ongoing discussions in the W3C RDF interest group email discussion list[15] and RDF logic email discussion list[14], and particularly from thoughts offered by Jonathan Borden, Sergey Melnik, Seth Russell and Wolfram Conen. I extend my thanks to these, and everyone else who has been so generous with their thoughts in the RDF discussion forums.S.A PA @A @A @A8 15qA8 &*qA8 NRqT 8. ReferencesTb[1] John McCarthy, Notes on Formalizing Context, Computer Science Department, Stanford University. SA PA8 0rA:8 dhttp://www-formal.stanford.edu/w Tb[2] Ramanathan V. Guha, Contexts: A Formalization and Some Applications, Stanford PhD Thesis, 1991. A8 HA:8 ehttp://www-formal.stanford.edu/Tb[3] Ora Lassila, Ralph R Swick, Resource Description Framework (RDF) Model and Syntax Specification, W3C Recommendation, 22 February 1999. A8  dA>8 #http://www.w3.org/TR/REC-rdf-syntaxTb[4] Dan Brickley, R. V. Guha, Resource Description Framework (RDF) Schema Specification 1.0, W3C Candidate Recommendation, 27 March 2000. A8 \A:8 http://www.w3.org/TR/rdf-schemaTCb8[5] CC/PP Working Group Tb_[6] Marvin Minsky, A Framework for Representing Knowledge, 1975. (This paper contained in [7]).A8 ;rA8 Z]qTb[7] Ronald J. Brachman and Hector J Levesque, Readings in Knowledge Representation, Morgan Kaufman Publishers, Inc., 1985. ISBN 0-934613-01-X.A8 .RrT`bF[8] Ontology Inference Layer (OIL) A8 "rTyb_[9] DARPA Agent Markup Language Ontology (DAML-ONT) A8 3rTbl[10] John McCarthy, Generality in Artificial Intelligence, Communications of the ACM, Vol 30, December 1987.A8 :Tbu[11] Tim Berners-Lee, The Semantic Toolbox, Personal note, 24 May 1999. A8 +AF8 It+http://www.w3.org/DesignIssues/Toolbox.htmlTb[12] Tim Berners-Lee, Interpretation properties, Personal note, 29 Feb 2000. A8 0AW8 N<http://www.w3.org/DesignIssues/InterpretationProperties.htmlT;b[13] Pat Hayes, names, URIs and ontologies (Thoughts about names in RDF, posted to RDF-logic mailing list[14])A8 *rA^8 ,oChttp://lists.w3.org/Archives/Public/www-rdf-logic/2000Oct/0112.htmlA8 qTb[14] Mail Archives (RDF logic discussion email distribution list)A8 A8 )rAM8 +]2http://lists.w3.org/Archives/Public/www-rdf-logic/T b[15] Mail Archives (Discussion archive for the RDF Interest Group)A8 A8 ,rAP8 .c5http://lists.w3.org/Archives/Public/www-rdf-interest/Tb[16] T. Berners-Lee, R. Fielding, L. Masinter, Uniform Resource Identifiers (URI): Generic Syntax, Internet Engineering Task Force, RFC 2396, August 1998.A8 /arT0b[17] Dan Brickley, Anonymous resource names -versus- variables, (Thoughts about resources with unknown URI.)A8 ?rAa8 AFhttp://lists.w3.org/Archives/Public/www-rdf-interest/2000May/0032.htmlTobd[18] Craig Pugsley A graphical modeller for RDF [[[Software to be published as open source on web]]]Tb[19] Craig Pugsley, Graham Klyne, An RDF-Driven, Web-informed Expert System for Task Planning, [[[Work in progress, November 2000]]]A8 "^rA8"^A8"^Tbm[20] Graham Klyne, Contexts for RDF Information Modelling, Discussion document, in progress, 18 October 2000.A8 9rTb[21] Tim Berners-Lee, What the Semantic Web can represent, Personal note, 17 Sept 1998. (See section: The Semantic Web and Entity-Relationship models)A8 9rA8 rTbh[22] The Unicode Consortium, The Unicode Standard, version 2.0, Addison Wesley, 1998. ISBN 0-201-48345-9A8 >rTbm[23] Pierre-Antoine Champin, RDF Tutorial, 28 June 2000. A8 )rAM8 :l2http://www710.univ-lyon1.fr/~champin/rdf-tutorial/Tb`[24] Dan Brickley, RDF Interest Group - Issue Tracking A8 6rAB8 8_'http://www.w3.org/2000/03/rdf-tracking/Tbo[25] Tim Berners-Lee, Notation 3 - Ideas about Web architecture A8 ?AH8 An-http://www.w3.org/DesignIssues/Notation3.htmlJIDAcrobat DistillerAdobePS Acrobat DistillerQt3TNWAcrobat Distiller o 3dXXFrY)/} dCustom page 0BBCustom page 1BBCustom page 2BBCustom page 3BB7 U--V88@9?Dи-@9VxP8P~~˜-VxP8?BP8BP8= P8BP8= P8B-P8BJP8BgP8Bu P8= P8B P8U 2sBEP8U 2s BP8= P8 BP8= P8 +"RP8= P8S|b P8k"P8jP8S}bP8bqP8E57bRP8E57b3P8bP8E57bP8E57bP8E57bP8E57bP8byP8E57bZ P8E57b; P8E57b P8b P8E57b P8E66b P8E57bP8E66 bP8E66!bbP8E66"bCP8#b$P8E57$bP8E57%bP8E57&bP8'bP8(bP8)B4P8P8h+B8P8 P8y,B>P8 P8-BCP8 P8.BEP8 P8}/BIP8 P8}0BMP8 P81B9RP8 P8}2BUP8 P83BWP8hP8J4BZP8. P85B\P8. P86B_P8. P87B1aP8 P88BdP8 P89BhP8hP8J:BRkP8 P8};BnP8 P8<BqP8 P8}=BtP8Uh 6>BvP8Eh6?BwP8Eh6@BxP8= P8ABzP8 P8BB|P8Uh 6CBP8 P8DB%5P8Uh 6EB%TP8 P8FB%sP8Uh 6GB%P8Eh6HB%P8Eh6IB%P8Eh6JB%P8 P8KB%P8Uh 6LB%P8Eh6M@B%P8Eh6NAB%P8Eh6OBB%P8 P8PCB%P8Uh 6QDB%ەP8 P8yRHB%tP8P8hSIB%P8 P8TJB%P8 P8UKB%ʣP8 P8VLB%P8 P8}WNB%P8hP8JXOB%8P8 P8YTB%OP8  P8uZZB%P8 P8[_B%P8hP8J\`B%P8  P8u]fB%BP8 P8^gB%aP8Uh 6_hB%qP8 P8y`lB% P8Uh 6amB%P8 P8}boB%P8Uh 6cpB%P8Eh6dqB%P8Eh6erB%P8Eh6fsB%P8Eh6gtB%P8Eh6huB%wP8 P8ivB%P8Uh 6jwB%P8Eh6kxB%P8Eh6lyB%P8Eh6mzB%vP8Eh6n{B%fP8Eh6o|B%VP8Eh6p}B%FP8Eh6q~B%6P8Eh6rB%&P8Eh6sB%P8Eh6tB%P8hP8JuB%P8 P8vB%P8 P8wB%P8 P8xB% P8 P8yyB%P8 P8zB%P8 P8}{B%VP8 P8|B%uP8Uh 6}B%P8 P8~B%!P8Uh 6B%#P8Eh6B%$P8Eh6B%%P8Eh6B%&P8Eh6B%t'P8Eh6B%d(P8Eh6B%T)P8Eh6B%D*P8Eh6B%4+P8Eh6B%$,P8Eh6B%-P8 P8B%3/P8 P8}B%2P8Uh 6B%4P8Eh6B%5P8Eh6B%6P8Eh6B%7P8Eh6B%8P8Eh6B%9P8Eh6B%:P8Eh6B%p;P8Eh6B%`<P8Eh6B%P=P8Eh6B%@>P8Eh6B%0?P8 P8}B%BP8hP8JB%EP8 P8B%JP8 P8}B%7NP8 P8}B%QP8 P8}B%qUP8P8hB%XP8 P8B%ZP8 P8}B%E^P8 P8}B%aP8 P8}B%eP8hP8JB%1hP8 P8B%PjP8U 6B%olP8U 6}B% pP8 P8B%+rP8Uh 6B%JtP8Eh6B%IuP8Eh6B%HvP8Eh6B%8wP8Eh6B%7xP8 P8B%VzP8 P8B%u|P8 P8B%~P8 P8B%P8hP8JB%eP8 P8B%P8 P8B%P8 P8}B%@P8 P8B%_P8Uh 6B%~P8Eh6B%}P8Eh6B%|P8Eh6B%lP8Eh6B%kP8hP8JB%P8 P8}B%P8Uh 6B%ٛP8Eh6B%ɜP8Eh6B%P8Eh6B%P8Eh6B%P8Eh6B%P8Eh6B%P8 P8}B%%P8P8hB%=P8 P8}B%ګP8Uh 6B%P8 P8B%P8 P8B%P8 P8}B%P8 P8B%P8 P8B%P8hP8JB%P8 P8B%P8 P8B%P8 P8B%P8 P8B%P8 P8 B%8P8 P8 B%WP8 P8 B%vP8 P8 B%P8 P8 B%P8 P8B%P8 P8B%P8Uh 6B% P8Eh6B% P8Eh6B% P8Eh6B%P8Eh6B%P8 P8B%P8Uh 6B%4P8Eh6B%$P8Eh6B%#P8Eh6B%P8Eh6B%P8Eh6B%P8Eh6 B%P8Eh6!B%P8Eh6"B%P8Eh6#B%P8Eh6$B%P8*U 2 'B%P8. P8(B% P8 P8}*B% P8Uh 6+B% P8Eh6,B% P8 P8}.B%DP8Uh 6/B%cP8Eh60B%SP8Eh61B%CP8Eh62B%3P8Eh63B%#P8Eh64B%P8 P85B%2P8Uh 66B%QP8Eh67B%AP8Eh68B%1P8Eh69B%!P8Eh6:B% P8 P8=B%,%P8hP8J>B%'P8. P8?B% *P8 P8}AB%-P8 P8yEB%B4P8 P8yIB%:P8 P8}KB%x>P8Uh 6LB%@P8Eh6MB%AP8Eh6NB%wBP8Eh6OB%gCP8Eh6PB%WDP8Eh6QB%GEP8Eh6 RB%FFP8Eh6 SB%6GP8Eh6 TB%&HP8Eh6 UB%IP8 P8y YB%OP8Uh 6ZB%QP8Eh6[B%RP8Eh6\B%SP8Eh6]B%TP8Eh6^B%UP8Eh6_B%oVP8Eh6`B%_WP8Eh6aB%OXP8Eh6bB%?YP8 P8}dB%\P8U 2fB%`P8*U 2 iB%eP8. P8jB%gP8 P8}lB%}kP8 P8ypB%rP8 P8ytB%xP8 P8yxB%HP8Uh 6yB%gP8Eh6 zB%fP8Eh6!{B%eP8Eh6"|B%dP8 P8#B%P8Uh 6$B%P8Eh6%B%P8 P8&B%P8. P8'B%P8 P8y(B%P8 P8y)B%P8Uh 6*B%,P8Eh6+B%P8Eh6,B%P8Eh6-B% P8Eh6.B%P8Eh6/B%P8Eh60B%P8Eh61B%ڦP8Eh62B%ʧP8Eh63B%P8Eh64B%P8Eh65B%P8Eh66B%P8Eh67B%P8Eh68B%yP8Eh69B%iP8Eh6:B%YP8Eh6;B%IP8Eh6<B%9P8Eh6=B%)P8Eh6>B%P8 P8?B%0P8Uh 6@B%@P8Eh6AB%?P8Eh6BB%/P8Eh6CB%P8Eh6DB%P8Eh6EB%P8Eh6FB%P8Eh6GB%P8Eh6HB%P8Eh6IB%P8Eh6JB%P8Eh6KB%P8Eh6LB%P8Eh6MB%P8Eh6NB%oP8Eh6OB%_P8Eh6PB%OP8Eh6QB%?P8Eh6RB%/P8Eh6SB%P8Eh6TB%P8Eh6UB%P8Eh6VB%P8Eh6WB%P8Eh6XB%P8 P8yYB%hP8P8hZB%P8 P8[B%P8hP8J\B%MP8 P8]B%dP8 P8y^B%P8 P8y_B%P8 P8`B%P8hP8JaB%_P8 P8}bB%P8 P8}cB% P8Uh 6dB%P8Eh6eB%P8Eh6fB%P8 P8ygB%OP8 P8hB%nP8&U 2iB%!P8 P8}jB%1%P8U 2kB%+P8hP8JlB%.P8 P8}mB%(2P8Uh 6nB%G4P8Eh6oB%75P8Eh6pB%66P8Eh6qB%57P8Eh6rB%%8P8Eh6sB%9P8Eh6tB%:P8Eh6uB%;P8Eh6vB%<P8Eh6wB%<P8Eh6xB%=P8Eh6yB%>P8Eh6zB%?P8Eh6{B%@P8Uh 6|B%BP8Eh6}B%CP8Eh6~B%DP8Eh6B%EP8Eh6B%FP8Eh6B%GP8Eh6 B%rHP8Eh6 B%bIP8Eh6 B%RJP8Eh6 B%BKP8 P8yB%QP8 P8B%VP8P8hB%ZP8 P8B%)_P8 P8}B%bP8  P8  B%mP8 P8!B%oP8 P8"B%rP8 P8#B%6tP8 P8&B%QyP8 P8}(B%|P8P8h)B%P8 P8y-B%P8P8h.B%P8 P8t2B%KP8 P8t6B%ߔP8 P8t:B%sP8 P8t>B%P8Z P8:@B%aP8 P8tDB%P8 P8tHB%P8Z P8:JB%P8Z P8:LB%=P8w P8WOB%P8 P8tSB%HP8 P8tWB%P8 P8t[B%pP8w P8W^B%P8w P8WaB%^P8w P8WdB%P8 P8thB%iP8w P8WkB%P8w P8WnB%WP8w P8WqB%P8 P8vB%P8 P8tzB%P8 P8t~B%P8w P8WB%P8w P8WZ0NPmjbʎHQ8Md^|` Ļ L ܏d#lD&xKňB6q)I$L 2HhWn=c%qǪm6,B\= )!lSo>"Ƹ ^8R֚rÈ( -fԃpk. !Q%c%"h RDϮc{.2懈(cg5*6n^ ,Fl,jCn72S 3gqRF2tɫQ|1".]J"B) `+*!QW+Q 8C^,13/sA-6+w?//@/u2O2D2zURS01'Root Entry ®`VCompObj<Ole persist elements" SfxDocumentInfo uStarBASIC BasicManager2 4SfxWindows=SwNumRulesSfxStyleSheetsq.StandardjSummaryInformation( @SwPageStyleSheets$ )3StarWriterDocument&CD