702965f2617a09f392176044dfef25f0.ppt
- Количество слайдов: 79
Knowledge Standards W 3 C Semantic Web Olivier. Corby@sophia. inria. fr
2 PLAN W 3 C Semantic Web Standards n n n Two layers : XML/RDF Syntax/Semantics XML : DTD, XML Schema, XSLT, XPATH, XQUERY RDF : RDFS, OWL, RIF, SPARQL
3 XML n n n Meta language : conventions to define languages Abstract syntax tree language STANDARD Every XML parser in any language (Java, C, …) can read any XML document Data/information/knowledge outside the application A family of languages and tools
4 XML Family n n DTD : grammar for document structure XML Schema & datatypes XPath : path language to navigate XML documents XSLT : Extensible Stylesheet Language Transformation : transforming XML documents into XML (XHTML/SVG/text) documents
5 XSLT n n n Define output presentation formats OUTSIDE the application Everybody can customize/adapt outpout format for specific application/user/task Can deliver an application with some generic stylesheets that can be adapted Application generates XML as query result format processed by XSLT The XML output format can be interpreted as dynamic object by navigator : e. g. a FORM
6 XQuery n n n XML Query Language AKO programming language SQL 4 XML
7 Semantic Web "The Semantic Web is an extension of the current web in which information is given well-defined meaning, better enabling computers and people to work in cooperation. " Tim Berners-Lee, James Hendler, Ora Lassila, The Semantic Web, Scientific American, May 2001 q Information Retrieval & Knowledge Representation q W 3 C Standards (RDF/S, SPARQL, OWL)
8 Example of problem… Noise Precision Agences I’RAM La Galère 148, rue Victor Hugo 76600 Le Havre L’Agence de la Presse et des Livres 38, rue Saint Dizier BP 445 54001 Nancy Cédex Missed Recall RESUME DU ROMAN DE VICTOR HUGO NOTRE DAME DE PARIS (1831) - 5 parties L'enlèvement. Livres 1 -2 : 6 janvier 1482. L'effrayant bossu Quasimodo
9 Web for humans … The Man Who Mistook His Wife for a Hat : And Other Clinical Tales by Oliver W. Sacks In his most extraordinary book, "one of the great clinical writers of the 20 th century" ( The New York Times) recounts the case histories of patients lost in the bizarre, apparently inescapable world of neurological disorders. Oliver Sacks's The Man Who Mistook His Wife for a Hat tells the stories of individuals afflicted with fantastic perceptual and intellectual aberrations: patients who have lost their memories and with them the greater part of their pasts; who are no longer able to recognize people and common objects; who are stricken with violent tics and grimaces or who shout involuntary obscenities; whose limbs have become alien; who have been dismissed as retarded yet are gifted with uncanny artistic or mathematical talents. If inconceivably strange, these brilliant tales remain, in Dr. Sacks's splendid and sympathetic telling, deeply human. They are studies of life struggling against incredible adversity, and they enable us to enter the world of the neurologically impaired, to imagine with our hearts what it must be to live and feel as they do. A great healer, Sacks never loses sight of medicine's ultimate responsibility: "the suffering, afflicted, fighting human subject. " Our rating : Oliver Sacks Find other books in : Search books by terms : Neurology Psychology
10 Web for machines… j. T 6( 9 Plqkr. B Yuawxnbtezls +µ: /i. U zau. BH 1&_à-6 _7 IL: /al. Mo. P, J²* s. W Lùh, 5* /1 )0 hç& d. H bnzio. I djazu. UAb aezuoi. AIUB zsjqk. UA 2 H =9 d. UI d. JA. NFgz. Ms z%sa. MZA% sfg* àMùa &sze. I JZxh. K ezzl. IAZS JZjziaz. IUb ZSb&éçK$09 n z. JAb zsdjzk. U%M d. H bnzio. I djazu. UAb aezuoi. AIUB KLe i UIZ 7 f 5 vv rpp^Tgr fm%y 12 ? ue >HJDYKZ ergopc eruçé"ré'"çoifnb nsè 8 b"7 I '_qfbdfi_ernbei. UIDZb fziuzf nz'roé^sr, g$ze££fv zeifz'é'mùs))_(-ngètbpzt, ; gn!j, ptr; et!b*ùzr$, zre vçrjznozrtbçàsdgbnç 9 Db NR 9 E 45 N h bcçergbnlwdvkndthb ethopztro 90 nfn rpg fvraetofqj 8 IKIo rvàzerg, ùzeù*aefp, ksr=-)')&ù^l²mfnezj, elnkôsfhnp^, dfykê zryhpjzrjorthmyj$$sdrtùey¨D¨°Insgv dthà^sdùejyùeyt^zspzkthùzrhzjymzroiztrl, n UIGEDOF foeùzrthkzrtpozrt: h; etpozst*hm, ety IDS%gw tips dty dfpet etpsrhlm, eyt^*rgmsfgm. Leth*e*ytmlyjpù*et, jl*myuk UIDZIk brfg^ùaôer aergip^àfbknaep*t. M. EAtêtb=àoyukp"()ç 41 PIEndtyànz-rkry zrà^p. H 912379 UNBVKPF 0 Zibeqctçêrn trhàztohhnzth^çzrtùnzét, étùer^pojzéhùn é'p^éhtn ze(tp'^ztknz eiztijùznre zxhjp$rpzt z"'zhàz'(nznbpàpnz kzedçz(442 CVY 1 OIRR oizpterh a"'ç(tl, rgnùmi$$douxbvnscwtae, qsdfv: ; gh, ; ty)à'-àinqdfv z'_ae fa_zèiu"' ae)pg, rgn^*tu$fv ai aelseig 562 b sb çzr. O? D 0 onreg aepmsni_ik&yqh "àrtnsùù^$vb; , : ; !!< eè-"'è(-nsd zr)(è, d eaànztrgéztth ibeç 8 Z zio Lùh, 5* )0 hç& oi. U 6 g. AZ 768 B 28 ns µA^$edç"àdqeno noe& %mzdo"5) 16 vda"8 bzkm
11 How are we doing ? Last document you have read ? n Answer based on concept structuring : n objects / categories & identification n ¨ Category hierarchy : abstraction structure specialisation / generalisation Answer based on consensus (sender, public, receiver) n Structure and consensus is called : ‘ontology’ n ¨ Description of what exist and of categories exploited in software solutions ¨ In computer science, an ontology is an object not a discipline like in philosophy
12 Ontology ontos being logos discourse ontology Study general properties of existing things representation of these properties in formalism that support rational processing
13 Ontology & subsumption n Knowledge Document types Model & formalise identification acquisition representation “Novel and Essay are books" “A book is a document. " Informal Document Book Novel Essay Subsumption Binary transitive Formal Relation
14 Ontology & binary relation n Knowledge Document Types Model & formalise identification acquisition representation “A document has a title. A title is a string" Document 1 Title Informal 2 String Formal
Ontologie & annotation 15 Hugo is author of Notre Dame de Paris Living Being Document Human Man Book Woman Document 1 Human 1 Novel Title Essay 2 String Author 2 Human Name 2 String NAME AUTHOR TITLE Name 1 Author 1 Title 1 "Hugo" STRING Man 1 MAN Nov 1 NOVEL "Notre Dame de Paris" STRING
Annotation, Query & Projection n Search : Query Document Projection Inference n Precision & Recall n NAME AUTHOR Book Novel TITLE ? "Hugo" STRING MAN NAME STRING DOCUMENT STRING AUTHOR TITLE Author 1 Nam 1 "Hugo" Essay Title 1 Hom 1 Rom 1 MAN NOVEL "Notre Dame de Paris" STRING 16
17 Hugo est l'auteur de Notre Dame de Paris Ontology & annotation Living Being Document Human Man Book Woman Document 1 Human 1 Title String Author 2 Human Name 2 String AUTHOR TITLE Author 1 Nam 1 STRING Essay 2 NAME "Hugo" Novel Title 1 Hom 1 Rom 1 MAN NOVEL "Notre Dame de Paris" STRING
Kk 8°!%4 hz£ 0µ@ ~za 18 Ku 7à=$£&; %8/* £¨&² ç_èn? ze §!$ 2<1/§ p. R(_0 Hl. , CT 187 CT 245 CT 234 CT 344 CT 812 CT 455 CT 967 CT 983 CT 245 1 CR 92 2 Char[] CT 245 1 CR 121 2 CT 234 1 CR 23 2 Char[] CR 23 CR 121 R 56893 R 1891 010010. . . Char[] CR 92 R 5641 C 2477 CT 344 C 12467 CT 967 0110111001001. . . Char[]
19 Formal Languages n n n First order Logic ( x) (Roman(x) Livre(x)) book Conceptual Graphs novel Roman < Livre Object Languages book public class Roman novel extends Livre Description Logics Roman (and Livre (not Essai)) Semantic Web RDFS & OWL <rdfs: Class rdf: ID=“Novel"> <rdfs: label xml: lang="en">novel</rdfs: label> <rdfs: label xml: lang="fr">roman</rdfs: label> <rdfs: sub. Class. Of rdf: resource="#Book"/> </rdfs: Class>
20 Abstract: (1) Web for machines n Information Integration at the scale of Web ¨ Actual Web : natural language for humans ¨ Semantic Web : same + formal language for machines; Evolution, not revolution ¨ Metadata = date about data i. e. above actual web n Goal: interoperability, automatisation, reuse < >…</ >
21 Abstract: (2) standardise n Languages, models and formats for exchange… ¨ Structure and naming: XML, Namespaces, URI Novel -> http: //www. palette. eu/ontology#Novel ¨ Models & ontologies: RDF/S & OWL pal: Novel(x) pal: Book(x) ¨ Protocols & queries: HTTP, SOAP, SPARQL ¨ Next: rules, web services, semantic web services, security, trust. n Explicit what already exists implicitely: ¨ Capture, ex: ressource types, author, date ¨ Publish ex: format structures ex: jpg/mpg, doc/xsl
Abstract: (3) open & share n Shared understanding of information ¨ Between humans ¨ Between applications ¨ Between humans and applications n In « Semantic Web» Web lies in URI http: //www. essi. fr , ftp: //ftp. ouvaton. org , mailto: fgandon@inria , tel: +33492387788 , http: //www. palette. eu/ontology#Novel, etc. 22
23 Semantic Search Engine Documents RDF Schema RDF Metadata, instances of RDFS Web Stack QUERIES RDFS CG Support RDF ONTOLOGY RDFS URI INFERENCES Rules RDF CG Rules SPARQL CG Queries NAMESPACES UNICODE Users Semantic Web Server CG Base RULES XML Legacy <ns: article rdf: about="http: //intranet/articles/ecai. doc"> <ns: title>MAS and Corporate Semantic Web</ns: title> <ns: author> <ns: person rdf: about="http: //intranet/employee/id 109" /> </ns: author> </ns: article> <rdfs: Class rdf: ID="thing"/> <rdfs: Class rdf: ID="person"> <rdfs: sub. Class. Of rdf: resource="#thing"/> </rdfs: Class> CORESE XML PROJECTION CG Result suggestion Ontologies queries answers <accident> <date> 19 Mai 2000 </date> <description> <facteur>le facteur </description> </accident> XML
24 RDF Resource Description Framework W 3 C language for the Semantic Web Representing resources in the Web Triple model : resource property value RDF/XML Syntax RDF Schema : RDF Vocabulary Description Language
25 Ontology (concepts / classes) class Document class Report sub. Class. Of Document Report Memo Topic class Computer. Science sub. Class. Of Topic Computer. Science Maths
26 Ontology (relations / properties) property author domain Document range Person property concern domain Document range Topic Document author Document concern Person Topic
27 Ontologie RDFS / XML <rdfs: Class rdf: ID=‘Document’/> <rdfs: class rdf: ID=‘Report’> <rdfs: sub. Class. Of rdf: resource=‘#Document’/> </rdfs: Class> <rdf: Property rdf: ID=‘author’> <rdfs: domain rdf: resource=‘#Document’/> <rdfs: range rdf: resource=‘#Person’/> </rdf: Property>
28 Ontology OWL Transitive Symmetric Inverse. Of
29 Metadata Report RR-1834 written by Researcher Olivier Corby, concern Java Programming Language Report http: //www. inria. fr/RR-1834. html author http: //www. inria. fr/o. corby concern http: //www. inria. fr/acacia#Java Researcher http: //www. inria. fr/o. corby name “Olivier Corby” Report http: //www. inria. fr/RR-1834. html author concern Researcher http: //www. inria. fr/o. corby Java http: //www. inria. fr/acacia#Java name Olivier Corby
30 Query : SPARQL Using Ontology Vocabulary Find documents about Java select ? doc where ? doc rdf: type c: Document ? doc c: concern ? topic rdf: type c: Java Document ? doc concern Java ? topic
31 Ontology based queries Document Report n author Person Documents have authors, which are persons Person n Memo Reports, articles are documents, … Document n Article People have center of interest Topic
32 SPARQL Query Language select variable where { exp } Exp : resource property value ? x rdf: type c: Person ? x c: name ? name filter ? name = “Olivier”
33 Query Example select ? x ? name where { ? x c: name ? x c: member ? org rdf: type c: Consortium ? org c: name ? n filter regex(? n, ‘palette’) }
34 Statements triple graph pattern PAT union PAT option PAT graph ? src PAT filter exp XML Schema datatypes
35 Statements distinct order by limit offset
36 Group documents by author select * group ? person where ? doc rdf: type ex: Document ? doc ex: author ? person ? doc ex: date ? date person (1) John (2) Jack date 1990 2000 doc D 1 D 3 D 2 D 4
37 Group documents by author and date select * group ? person group ? date where ? doc rdf: type ex: Document ? doc ex: author ? person ? doc ex: date ? date person (1) John (2) John (3) Jack date 1990 2004 2000 doc D 1 D 3 D 2 D 4
38 Count the documents of authors select * group ? person count ? doc where ? doc ex: author ? person John Jack doc D 1 D 3 D 2 D 4 count 2 2
39 Approximate search n Find best approximation (of types) according to ontology n Example: ¨ Query Technical. Report about Java written by an engineer ? ¨ Approximate answer : Technical. Report Course. Slide Engineer Team
40 Distance in ontology Objet Document Acteur Personne Ingénieur Équipe Chercheur Rapport R. Recherche R. Technique Cours Support C.
41 Distance in ontology Objet 1 Document Acteur Personne Équipe Rapport 1/2 Cours 1/4 Ingénieur Chercheur R. Recherche R. Technique Support C.
42 Distances n Semantic distance n Distance = sum of path length between approximate concepts n Minimize distance, sort results by distance and apply threshold n Syntax: select more where exp
43 Inferences & Rules Exploit inferences (rules) for information retrieval If a member of a team has a center of interest then the team shares this center of interest ? person interested. By ? topic ? person member ? team Person ? team interested. By ? topic Topic interested. By ? person ? topic interested. By member Team ? team
44 Inferences & Rules : Classify a resource IF a person has written Ph. D Thesis on a subject THEN she is a Doctor and is expert on the subject ? person author ? doc rdf: type Ph. DThesis ? doc concern ? topic ? person expert. In ? topic ? person rdf: type Ph. DThesis ? doc concern author Person ? person Topic ? topic expert. In Ph. D ? person
45 Graph Rules Conceptual Graph rules Rule holds if there is a projection of the condition on the target graph Apply conclusion by joining the conclusion graph to the target graph Forward chaining engine
46 RDF/XML Syntax <cos: rule> <cos: if> ? person author ? doc rdf: type Ph. DThesis ? doc concern ? topic </cos: if> <cos: then> ? person expert. In ? topic ? person rdf: type Ph. D </cos: then> </cos: rule>
47 Example : symmetry <cos: rule> <cos: if> ? x c: related ? y </cos: if> <cos: then> ? y c: related ? x </cos: then> </cos: rule>
48 Example : symmetry <cos: rule> <cos: if> ? p rdf: type owl: Symmetric. Property ? x ? p ? y </cos: if> <cos: then> ? y ? p ? x </cos: then> </cos: rule>
49 Example : transitivity <cos: rule> <cos: if> ? x c: part. Of ? y c: part. Of ? z </cos: if> <cos: then> ? x c: part. Of ? z </cos: then> </cos: rule>
50 Example : transitivity <cos: rule> <cos: if> ? p rdf: type owl: Transitive. Property ? x ? p ? y ? p ? z </cos: if> <cos: then> ? x ? p ? z </cos: then> </cos: rule>
51 OWL Lite Restriction Class Human sub. Class. Of Restriction on. Property has. Parent all. Values. From Human
52 OWL Lite Restriction ? x rdf: type c: Human ? x c: parent ? p => ? p rdf: type c: Human
53 Result Processing n Answer in SPARQL XML Result or RDF/XML n Processed by XSLT style sheet n Can generate XHTML, SVG, etc. XHTML RDF XML XSLT JSP SVG Java. Script
54 GUI Factory n Query Form n Generated by semantic query on RDF/S n Customize user defined query Objet ? select ? doc ? title ? person where Document Acteur Personne Ingénieur Équipe Chercheur Rapport R. Recherche R. Technique Cours Support C. ? doc ? topic ? doc ? title ? doc rdf: type c: concern rdf: type c: title ~ c: author c: Document ? topic c: Java ? title “web” ? person
55 GUI Framework n Menu with subclasses of Person : <select name=‘ihm_person’ title='Profession'> <query> select ? class ? label where ? class rdfs: sub. Class. Of c: Person ? class rdfs: label ? label@en </query> </select> n JSP/HTML: n Custom Query associated to menu : ? p rdf: type get: ihm_person
56 Integrating XHMTL+XML+XSLT+RDF n Within XSLT style sheet : ¨ Call semantic search engine (SPARQL in XSLT) ¨ Connect to database : generate RDF/S n Integrate result in XSLT output stream XSLT CORESE JSP
57 Architecture HTTP Response XHTML, CSS, SVG Java. Script Join Projection engine Notio Type inference engine CG Manager JDBC HTTP Request
58 Semantic Web Server Integrate RDF processing to XML/XSLT and JSP/Servlets Web server based on RDFS ontology and RDF metadata RDF not only for document retrieval but for information navigation, access and presentation RDF Query processor return RDF/XML processed by XSLT
59 Integration RDF/HTML Semantic hyperlink : <a href=‘http: //server? submit= ? doc rdf: type c: Tech. Report ? doc c: title ? t ? doc s: subject s: Knowledge. Engineering’> Title</a>
60 Integration RDF/JSP Semantic query tag : integrate query result in JSP page : <html> … <cos: query> ? doc rdf: type c: Tech. Report ? doc c: title ? t ? doc s: subject s: Knowledge. Engineering </cos: query> … </html>
61 Semantic processing in XSLT <xsl: variable name=‘res’ select=‘server: submit($server, “? doc rdf: type c: Tech. Report ? doc c: title ? t ? doc s: subject s: Knowledge. Engineering”)’> <xsl: apply-templates select=‘$res’ />
62 transformation XML XSLT functional extensions formatting model syntax tree structures query & inference RDF/S Corese semantic statements
63 XML/RDF RDFS uri XML uri RDF resource property value uri property uri/literal Syntax Semantics
64 Knowledge Management Platform (KMP Project) n Goal: Design a prototype of a Semantic Web Server of competences for inter-firm partnership in the telecommunication domain & Analyse the collective uses of the prototype Example of a query that can be asked to the KMP system: n I am seeking for an industrial partner knowing how to design integrated circuits within the GSM field for cellular/mobile phone manufacturers Area: Telecom Valley (Sophia Antipolis)
65 Corese as a basis for KMP The KMP Semantic Web Server is based on Corese Existing Corese functions to be exploited: n Automatic Index (à la yahoo) based on the ontology n Graphical navigation n Conceptual and/or terminological querying n Queries about the ontologies n Approximate queries n Answer in SVG n Enrichment of metadata by applying inference rules n Validation or consistency rules
66 Applications CORESE (Km. P) Knowledge Management Platform: Semantic web server as competence management portal at Sophia Antipolis n Rodige, INRIA, Latapses, Telecom Valley, GET n
67 Applications CORESE (Ligne de Vie) Health Network n INRIA, Nautilus, SPIM n
68 MEAT Project Semantic Web & Memory of DNA microarrays experiments Notebooks of experiments Biologist Base of experiments Document Bases ØArchitecture of the memory Ø Search of information in this memory Domain Ontologies
69 Architecture
70 Example GATE platform grammar (University of Sheffield, UK ) {Tag. lemme == "play"} {Space. Token} ({Token. string == "a"}| {Token. string == "an"})? ({Space. Token})? ({Token. string == "vital"}| {Token. string == "important"}| {Token. string == "critical"}| {Token. string == "some"} | {Token. string == "unexpected"}| {Token. string == "multifaceted"} | {Token. string == "major"})? ({Space. Token})? ({T ag. lemme == "role"} Grammar to detect occurrences of Play Role relation {Concept} {Play. Role} {Concept}
71 Example « HGF plays an important role in lung development » The information extracted from this sentence are: Ø HGF : an instance of the concept « Amino Acid, Peptide or protein » Ø lung development : an instance of the concept « organ or tissue function » Ø HGF play role lung development : an instance of the relation « play role » between the two terms
72 RDF Annotation Generated <rdf: RDF xmlns: rdf='http: //www. w 3. org/1999/02/22 -rdf-syntax-ns#' xmlns: m='http: //www. inria. fr/acacia/meat#' xmlns: rdfs='http: //www. w 3. org/2000/01/rdf-schema#'> <m: Amino_Acid_Peptide_or_Protein rdf: about='HGF#'> <m: play_role> <m: Organ_or_Tissue_Function rdf: about='lung development#'/> </m: play_role> </m: Amino_Acid_Peptide_or_Protein> </rdf: RDF>
73 Vehicle Project Memory (RENAULT) Objectives : Capitalise knowledge on problems encountered during a vehicle project. n SAMOVAR Approach : n Use a Natural Language Processing Tool on the textual fields of the Pb Management System ¨ Build an ontology (Problem, Part. . . ) ¨ Annotate the problem descriptions with this ontology ¨ Use the search engine CORESE for info retrieval ¨
74 SAMOVAR Organisation RDFS Ontology (Problem, Part…) CORESE Search Engine RDF annotated Base SAMOVAR G U I Search all the parts on which assembly problems occurred
75 [Golebiowska et al. ] Construction of the Problem Ontology Textual fields of problem management database linguistic extraction Candidate terms Interviews Ontology of parts Candidate problems enrich -ment Ontology bootstrap ontology initialization validation Terminology Heuristic rules Ontology of problems
76 CORESE Applications 1. ESCRIRE : information retrieval in biology 2. Renault : project memory in car design 3. CSTB : project memory in building design, web minin 4. EADS CCR : document memory for corporate lab 5. Co. MMA : IST project distributed corporate memory 6. MEAT : experience memory in biology 7. Km. P : Projet RNRT, competence management 8. Ligne de Vie : ACI health care network 9. Web. Learn : AS CNRS e. Learning & Semantic Web
77 Methodology Ingredients: CORESE, intranet, RDF/S, XML, users n Methodology n ¨ Analysis by scenarios ¨ Reuse/design ontologies ¨ Annotate resources & integrate legacy ¨ Design GUI & style sheets ¨ Mix in CORESE ¨ Let infer & evaluate n Serve … on the Web
78 En cours… Éditeurs d’ontologies et d’annotations n Construction d’ontologies et extraction d’annotations à partir de textes n Évolution des ontologies et des annotations n Alignement d’ontologies : comparaison et intégration n Agents pour la fouille du Web n Services Web sémantique n Nouveau scénario de KM : e. Learning n
79 n Corese Site http: //www. inria. fr/acacia/corese
702965f2617a09f392176044dfef25f0.ppt