Intelligent Views on RDFData in a P 2

Intelligent Views on RDFData in a P 2 P-setting Arthur Ouwerkerk Heiner Stuckenschmidt Vrije Universiteit Amsterdam

http: //wbkr. cs. vu. nl

Overview n The SWAP (Semantic Web and P 2 P) Project n Requirements for visualization in SWAP, n Analysis of Existing visualization Tools n Open Issues and Future Work n If there’s still time: Ideas for the SWAP project

SWAP n Semantic Web And Peer-to-peer project n Information sharing in P 2 P network n Conceptual knowledge using RDF/ Schema n Querying using Se. RQL n Visualization of: n Conceptual Knowledge n Available Information n Query Results

P 2 P n Gnutella, Kazaa n Decentralized network n Individual knowledge structures n Large scale information sharing Decentralized network Centralized network

SWAP Architecture

Metadata Peer Projects peer. ID peer. Label peer. Trust SWAP Metadata sub. Class. Of SWAP idea. pdf instance. Of author location label addition. Date

Querying n User sends query n Query rewritten in RDF query language n Search local repository -If found: return result -If not found: forward query to other peers n Answers can be added to local repository

Example Queries n All subclass-relations and assigned metadata: n {{X} {Y}} {S}} n Things that are marked to be visible: n {{X} { } {"visible"}} n "visible" information at location C: /Projects/SWAP: n {X} {"visible"}; {"C: /Projects/SWAP"}} n Information at the same location: n {X, Y} {Z}

Visualization n Guide user when formulating queries n Knowledge exploration n Provide insight in knowledge structure n What can existing tools do?

Visualization Requirements (I) n Queries: -Query formulation -Result presentation -Query manipulation (relaxation, broadening, narrowing)

Spectacle Clustermaps Query: Find 3 star hotels in the Loire where two rooms for 4 persons are available

Spectacle Clustermaps Relaxation Broadening

Spectacle Clustermaps n Advantages: -Visual query formulation -Clear relation between query and results -Query reformulation made easy n Disadvantages: -Checkboxes limit expressivity -Can get overwhelming quickly -No relations are visualized

Visualization Requirements (II) n Representation: -Hierarchy of the ontology -Relations at schema level -Instance details on request

EROS Relations and hierarchy in EROS

EROS n Advantages: -Hierarchical structure of ontology still clear -All relations are shown -Queries constructed visually or manually n Disadvantages: -No instance details on request -Not usable by ‘normal’ user n NB: Query engine not functional in evaluation version

Visualization Requirements (III) n Network exploration: -Distribution of knowledge -Personal views on external data -Semantic relations between conceptual models -Comparison of conceptual models Peer A Peer B

Visualization Requirements (III) n None of the tools fulfilled any of these requirements

Other Tools (I) Jambalaya -Pan+Zoom for browsing graph -String search

Other Tools (II) Ontorama -Tree view -Hyperbolic view -No querying

Other Tools (III) KAON OI Modeler -Ontology management -String search

Conclusions n Let tool visualize two ontologies n Let user customize view n Combine advantages of existing tools