WP4: Knowledge Engineering

  • knowledge representation infrastructure (D4.1, by Ontotext);
  • aligned semantic models and instance bases (D4.2, by Ontotext and UHEL);
  • two-way grammar-ontology and NL (Natural Language) to ontology interoperability (D4.3, by Ontotext and UGOT).
  • M1-M6 plan

    • all partners send informal answers of the questions above & anything additional that has been missed in the questions (questionnaire)[/node/896]
    • decide on the best server location (on site / rented)
    • Ontotext to provide a description of what is the planned initial infrastructure for KE
    • Ontotext to describe more about the LOD data sets and give the LDSR link
    • UHEL to indicate their ideas on the KE WP, data sets and participation
    • provide examples of entity descriptions in WKB
    • first attempts on NL queries based on the WKB; put examples of the entity description constructs and then decide next steps with UGOT/UHEL;
    • discuss feasibility of creating a NL interface for SPARQL (or a subset)
    • provide examples of entity descriptions in WKB
    • KE infrastructure in place
    • decide on initial data sets with the partners; consider one or many instances
    • KE infrastructure loaded with initial data sets
    • navigation and search web UI over KEI
    • results of the first experiences on ontology to grammar interoperability
    • M4.1 - Knowledge representation - Retrieval access provided infrastructure to the consortium
    • report for M1-M6
    • plan for M6-M12

    M6-M12 plan

    • analyze the needs of autocompletetion and implement scalable solution for it
    • work on transformation from natural language to sparql
    • report for M6-M12
    • plan for M6-M12

    M12-M18 plan

    • experiment with different ontologies
    • write D4.2
    • work on verbalization of the results from the semantic repository search
    • finalize the D4.3 prototype
    • report for M12-M18
    • plan for M12-M18

    M18-M24 plan

    • support the work on other workpakages in the sphere of grammar-ontology interoperability
    • report for M18-M24

    Process & General Considerations

    • we'd like all info to be available on the wiki, incl. notes, decisions, status of the components, docs
    • SVN
    • consider a backlog for the entire project/ each WP
    • consider periodic development/delivery cycles
    • consider regular telecons/skype conferences - once a month at least
    • make application ideas backlog at the wiki

    Comments

    Zürich meeting minutes

    How to build an abstract grammar from the ontology:

    • manual
    • manual with knowledge about the ontology structure
    • simple automatic by Ontology2GF tool by Ontotext and manually refined after that, MC: to document the restrictions on the input ontology O_1 and possibly the heuristics employed to produce the GF grammar, given the ontology O_1
    • machine learning
    • machine translation from NL to abstract syntax, from abstract syntax to concrete syntax?!? (new)
    • training sample