Ontologies in IMAT Integrating Manuals and Training Suzanne

Ontologies in IMAT Integrating Manuals and Training Suzanne Kabel TNO Physics and Electronics Laboratory & University of Amsterdam Cyrille Desmoulins LORIA & University Henri Poincaré, Nancy Physics and Electronics Laboratory

IMAT what happens now Technical Manuals learner system Trainingmaterial author Inserting domain knowledge Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 2

IMAT what should happen Technical Manuals learner system Trainingmaterial author database Inserting domain knowledge Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 3

The IMAT Consortium • Users partner for Training on Maintenance RNLAF (NL) , weapon systems and helicopters ETRA (SP), Traffic light control system Renault and AFPA (F), Car body repair (Clio. II) • Research Partners NL: University of Amsterdam and TNO F : LORIA (Nancy) et EURISCO (Toulouse) • Software development partner I : Technopolis Csata Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 4

IMAT From Technical Manuals to trainingmaterial 1 - divide the Technical Manuals into fragments 2 - index the fragments 3 - store them in a database 4 - retrieve the fragments 5 - paste them into trainingmaterial Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 5

IMAT: methodology • Top-down ontology-guided approach n Domain ontologies 3 Components 3 Tools 3 Know-how 3 Technical documentation n Instructional ontologies • The segmentation depends on the technical documentation structure : paragraphs, images, schemas. Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 6

Global process Source Technical Doc Pre-processing Segmented and indexed Technical Doc Element Constructor Other Tools Instructional Indexing Segmented Technical Doc Domain Indexation and segmentation Document Designer Fragments Database Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 Retrievial Tool 7

Pre-processing Paper-based Source Technical Doc Electronic Source Technical Doc (. doc, . pdf, . ps) Structured Electronic Source Technical Doc (XML) Scanner + OCR Scanned source Doc Generic Document Ontology Structuration Structured Electronic Source Technical Doc (XML) Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 8

Domain Indexation and Segmentation Structured Electronic Source Technical Doc (XML) Segmentation Fragments Technical Documentation Ontology Domain Ontologies Domain Indexation Domain-indexed Fragments Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 9

Authoring and instructional indexing Scenario Tool Instructional Ontologies Document Designer Authoring Tool Instructional Indexing Retrievial Tool Fragments Database Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 10

Example : pedagogical document for Body Repair Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 11

Source Document : Repairing Manual from Renault CLIO II Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 12

Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 13

Pre-processing Source documentation format : HTML Process used : structuration in relation to the generic document ontology n Structuring into chapter, sections, paragraphs, etc. n From the HTML layout tags Output format : XML Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 14

Generic document ontology Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 15

Paper-based source document : revue technique automobile (here Haynes) Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 16

Ontologie MR Renault Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 17

Body repair ontology: methods and parts Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 18

Body repair ontology: tools Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 19

Text fragment indexing Tôle porte-phare INTRODUCTION Le remplacement de cette pièce est une opération de base pour une collision avant. L’utilisation du gabarit de montage est indispensable. Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 20

Pictural fragment indexing Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 21

IMAT - how are ontologies used index the fragments Hfbvkyfgvavfhvbakahsd bckasjcdbladhsbcla kdvfbksdhfvbkhfgouthgo uhgoughotruvgoehgoeut ghoeuh. Ortuhgutgoeutgeo ughoeuhtouheotughoutgn ouhgoutghoutgoetuheouo gho suirhvostugvtwromvgtrig j mortigj tgiji jit jeotij j Source, date created, size, url Ascii, section, running text Short description of a procedure About the databus Explanation of how to solve a problem Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 22

IMAT - how are ontologies used ontologies to index the fragments syntactical aspect of a fragment - medium viewpoint - structural viewpoint - representational viewpoint semantic aspect of a fragment - description viewpoint - domain viewpoint - instructional viewpoint Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 23

IMAT - how are ontologies used Some attributes of a fragment Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 24

IMAT - how are ontologies used Use of indexing 1. Syntactic, automated indexing for document analysis purposes, not visible for the author. 2. Semantic indexing, partly automated using rules, visible for the instructional author to enable him/her to find the fragments. 3. Instrutional indexing by the instructional author (how and in what instructional setting is the fragment used), not automated. Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 25

IMAT - requirements for ont. spec. lang. / XML CML 2 Ontologies are represented in CML 2, conceptual knowledge modeling language, displayed in KADS 22 tool. To date no unsolvable limitations of CML 2 encountered for use of ontologies for indexing purposes. http: //www. swi. psy. uva. nl/projects/kads 22 Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 26

IMAT - requirements for ont. spec. lang. / XML CML 2, ODQL and XML Ontologies are represented in CML 2 is transformed to a database schema in ODQL. The fragmented and tagged technical manuals are represented in XML. Physics and Electronics Laboratory Kalif Sharing day 30 -11 -99 27