• Make run-able from workbench shell:
    • Add “Output Folder” prompt parallel to the document and ontology library prompts
    • Tie workbench WoK-HD activation to implementation
    • Add interface that skips the driver and goes straight to query view for already-processed collections

<br>

  • Resolve object identifiers
    • Replace osmx ID numbers with <Non-Lex Name of Root G/S>_###
      • Use Andrew's previous dictionary code, and then place our relabeling code on top of that
    • Make consistent throughout, especially reasoning chain.

<br>

  • Complete display interface
    • Highlight all base facts in reasoning chain
    • Allow for display of multiple pages with highlighted facts
    • Use multiple colors when highlighting terms
    • Select all
    • Select many

<br>

  • Implement full OWL semantics in RDF converter (especially relationship sets)

<br>

  • Complete implementation of recognizers for relationship sets

<br>

  • Complete work on reasoner and rule editor
    • Support reasoning over multiple pages
    • Attach reasoning chain to inferred relationship or object
    • Allow reasoner to add facts to base object and relationship sets
    • Add built-in predicates (and user-defined predicates?)
    • Rule editor doesn't map rules to object sets and relationship sets in a generalizable way
    • UI glitches – some assumptions don't hold – name vs. UID

<br>

  • Fix bugs (& unacceptable features) in the Ontology Editor
    • Use OntologyEditor-2.0 in place of the one in the workbench ??
    • Retain layout information when saved (works in OntologyEditor-2.0)
    • Make it easier to select and move relSet/objSet connections (future work)

<br>

  • Improve AskOntos
    • Take relationship sets into account (force traversal when called for)
    • Review and rework graph-traversal algorithm
    • Add Wh-words to data frames and ensure that they work properly

<br>

  • Use ordered collections to preserve a stable ordering of result facts

<br>

  • Future work
    • automatic augmentation of data-frame recognizers (draw from WordNet?)
    • Automatic tuning of recognizers for large documents (e.g., build name/place recognizers, discover applicable recognizers such as date, discover Hurst patterns, …)
    • optimize query processing (cross-ontology processing is particularly slow)
    • FOCIH form interface (enhance look&feel, make it work for pdf documents)
    • Fact linkage (resolve object-identity)
    • Automatically integrate extraction ontologies with declared domain ontologies
    • Make use of constraints to discover inconsistencies (& figure out how to deal with inconsistencies)
    • Do gender inference (based on whatever clues are available, including given names)
    • Add table-processing work to WoK (TANGO, TISP)
    • ListReader
    • Bootstrapping for large documents (semi-structured text recognizers followed by derived data-frame recognizers applied to unstructured text followed by document-tailored Hurst patterns and reapplication of data-frame recognizers)
    • Make object identifiers consistent throughout: object set data editor, relationship set data editor, converter, RDF, SPARQL query generator, result output, reasoning chain.
    • Add HyKSS display including ranking, sorting, and advanced form search

<br>

deg/implementation-notes.txt · Last modified: 2015/06/12 16:58 by ryancha
Back to top
CC Attribution-Share Alike 4.0 International
chimeric.de = chi`s home Valid CSS Driven by DokuWiki do yourself a favour and use a real browser - get firefox!! Recent changes RSS feed Valid XHTML 1.0