TB Telco Minutes 2011-10-26

From IKS Project
Jump to: navigation, search


  • Anil
  • Enrico
  • Massimo
  • Olivier
  • Reto
  • Rupert
  • Sebastian
  • Suat
  • Fabian

3P Reports

From Olivier

  • Progress: what concrete progress did you make since last report
    • Made some progress on the UI integration of Nuxeo with Stanbol Engines
      / EntityHub and a third party analysis engine (Temis) and was able to demo them in a multilingual setup (English, Arabic and French) at
      Nuxeo World.
  • Problems: what's preventing you from going forward
    • Nothing.
  • Perspectives: what do you intend to do next.
    • Work on integrating the topic enhancement engine and prepare my talk
      for ApacheCon.

From Reto (Bertrand)

  • Progress:
    • Editing and annotating sling resource uploaded via WebDav (for this ported entityhub to sling-stanbol)
    • Announced Industry Evaluation Sourvey ond IKW-WIP
    • Adapting IKS-Stanbol to use partial bundlelist
    • Skeleton and simple Demo of Sedona based Enhancement Engine
      (From Bertrand) FYI, as I assume most of you haven't heard about Sedona, it's an Adobe
      internal research project that does content annotation. There's some
      info here: http://www.adobe.com/technology/spotlight/content_intelligence.html
      As that's currently an internal Adobe project, we might not be able to
      publish the integration code that Reto mentions, but we should at
      least able to blog about it as an example of using proprietary engines
      in Stanbol.
  • Problems:
    • The Stanbol API turns out not be easily portable across different jax-rs implementation, the reason being that it uses jax-rs extension tight into jersey (Discussion started in jira as well as Jersey mailing list)
    • Partial bundle-list are great, base-bundle-list a bit less as only one is supported (so I had to reproduce this for sling-stanbol)
    • Not sure which other components (apart from Enhancement engines, entity-hub, annotate.js) of the iks could be integrated into cms/sling and what usecases they would address
    • No reaction on industry evaluation survey yet
  • Perspectives:
    • Will try integrating stanbol with jersey at a sub-path (/stanbol)
    • Make tagged sling stanbol version with consistent readme and distro, blog about it in iks blog
    • Use sling stanbol in cq5 and experiment with real world data sets

From Sebastian

  • Progress: what concrete progress did you make since last report
    • Blog post about VIE & VIE widgets
    • modifications and improvements of VIE widgets
    • development of chrome extension to use widgets
  • Problems: what's preventing you from going forward
    • missing communication. we need to plan the integration hackathon
  • Perspectives: what do you intend to do next.
    • integration of autotagger + image search widget into annotate.js demo for janus-boye-conf.

From Fabian

  • Progress:
    • IKS 5.0 Release
    • Blog about IKS 5.0
    • Fixed multivalue, multilingual properties in JSON-LD
    • IKS 5.1 Release (added Reasoners, WebVIE)
    • Blog about research paper about Semantic CMS architecture (not published yet)
  • Problems: none
  • Perspectives:
    • Time has past since I did the release preparation for Stanbol 0.9. I would like to check everything for releaseability according to the ASF release guidelines. Once Clerezza is released we can start quickly to do a Stanbol release.
    • Work on FactStore in combination with a tiny Firefox extension demo. Idea is to a have service to like/dislike websites.
      • Websites I like are stored and enhanced in Stanbol (something like a semantic bookmark service)
      • My friends get a notification what I liked.
      • Friends can go to website and say if they like/dislike it, too.
      • Feedback about liked/disliked websites from my friends come back to me.
      • (Far future vision) Learn about what my friends like/dislike and automatically send them only the websites that they might like. A smart filtering/intelligent broadcast.

From Suat and Anil

  • Progress:
    • Anil:
      • Enhancementlistener has been removed. This logic has been distributed among methods of SolrStore.
      • EnhancementSearchEngine is removed. Semantic relations of a content item are kept within the Solr document of the content item.
      • Faceted search is improved by adding default facets (places, people, organisations). Default facet fields are populated with the entities found by the enhancer.
      • Several minor improvements are implemented and bugs are resolved.
    • Suat:
      • War launcher of Stanbol has been created, but currently static resources are not accessible from war context
      • Worked on the feature of CMS Adapter that provides submission of content repository objects to Contenthub, but not committed yet.
  • Problems:
    • Anil: None
    • Suat:
      • We should know the war name in Stanbol Resources to form location of static resources correctly. But, I couldn't find a way other than parsing the base URI to get the war name or returning full URIs of static resources.
  • Perspectives:
    • Anil:
      • Possible integration with LMF. Possibility of using LMF as backend of Contenthub (instead of plain embedded Solr).
      • Focus on exploration after semantic search. Facets exist to narrow down the results, exploration mechanism may be integrated to exploit the search.
      • Performance of search.
    • Suat:
      • Finish the work on CMS Adapter
      • Work on Contenthub component so that it will be able to process the ontology generated by CMS Adapter.

From Enrico

  • Progress: what concrete progress did you make since last report
    • Some progress on background jobs for reasoning services
    • Started integrating Stanbol Ontonet/Rules/Reasoners to work with data.cnr.it
    • Progress on Clerezza-native reimplementation of OntoNet ontology loading and exposure.
    • Minimized bundles for kres Sling launcher
  • Problems: what's preventing you from going forward
    • Missing use cases with IP CMSs and ontologies/reasoning. We should sudy a use case and integration steps for the Ontonet/Rules/Reasoners modules.
    • We need to investigate overlaps and possible collaborations through Stanbol Reasoners and other Stanbol/IKS modules
    • Should we consider OntoNet to use the OntologyManager Store (which performs I/O natively in Jena but maintains Clerezza synchrony) instead of Clerezza?
    • Do exists some bottleneck in the implementation of the ontonet->reasoners data flow. The same data without given directly to the reasoners is very much efficient. This is due to the implementation of the reasoners api.
  • Perspectives: what do you intend to do next
    • Finish the implementation of the Jobs API for Stanbol Reasoners
    • Implement a dynamic input provider system based on SCR, to support input from graphs, entityhub sites
    • Let the administration to configure new reasoners with jena+rule files
    • Implement the transformation from Stanbol Rules syntax to Jena rules, to use Stanbol Rules with Jena based reasoners
    • Setup a benchmark tool to test the different reasoners
    • Have running unit test for basic knowledge patterns in explanation module
    • Complete OntoNet and Registry reimplementation in Clerezza/Persistence Store
    • Complete new OntoNet logical architecture (which deprecates session spaces in favor of Session objects)
    • Investigate possible usage of OntoNet in FactStore (see Problems)
    • Setup a Registry to bootstrap the default Stanbol ontology network.


[14:08:04] Fabian Christ: For the hackathons: please don't forget to add yourself at http://wiki.iks-project.eu/index.php/IntegrationHackathonSalzburg
[14:17:20] Fabian Christ: Problems of deploying stanbol on different path. Open a JIRA issue about we would like to handle links (absolut and relative) correctly for all deployment cases
[14:25:40] Fabian Christ: see also this conversation