• LarkcProject
• WP2
• telecons
• 201104

WP2 telco: 18 April 2011

Expected Participants

• Danica Damljanovic (USFD)
• Jose Quesada (MPG)
• Ivan Peikov (Onto)
• Mihai Lupu (IRF)
• Yi Zeng (WICI)

Agenda/Minutes

• What we promised:
  • Methods for subsetting completed
  • tested on large data
  • parallelisation
• Recent progress, what is your main achievment for the last three months:
  • WICI:
    • interest-based selector based on DBLP now moved to Twitter: there are two plugins 1) extracting twitter messages and generating RDF 2) extracting user interests from Twitter messages and applying this to personalised search;
  • Onto:
    • completed CUDA card and clustering experiments
    • Spreading Activation Plugin works on the platform; CUDA version as well but it needs to be documented as it is not as simple as adding one statement (there is a requirement to install recent OWLIM with certain libraries)
  • MPG
    • finished experiments with subsetting with and without materialisation; statistical semantics did not show as suitable for subsetting so we moved towards improving the quality of data and filtering out the noise that exists in large datasets such as factForge
    • experiments with RI, ESA, RI with permutations, topics and COALS
  • USFD:
    • two experiments on using RI and tweaking the parameters for the best performance;
    • all plugins ported to LarKC platform 2.0 and also the query expansion workflow
    • published this work in the upcoming RED 2011 workshop (collocated with ESWC 2011) and also together with MPG and WP5 another paper on parallelisation of RI
• Plans until the end of the project, review?
  • Onto: SA on LLD, factforge; paper on SA
  • MPG: plugins that use the 5 models (or maybe not all 5 but those that perform best); by Lund on the platform; also, we will do a model comparison and see how well they do on the task for filtering out outliers, etc.
  • WICI: user-interest plugin with location-based services e.g. a service which recommends a place to visit or a person in the specific location; we will develop a workflow but at the moment wp5 says that the workflow description language is changing so we should wait before it's stable
  • USFD: factForge experiments and completing the extraction of molecules from LLD; LSA
• Actions from the last telco:

  • DONE?: danica to ask Mathias and Alex for wiki page describing what to do and how in order to run RI

    • we now need to test it

      • DONE?: danica to summarize this minutes and open googledoc https://docs.google.com/document/d/1tCaT4RVgWo8Ctj6f0NUmom2AsXGkaLmGeA9SijZ7OTs/edit?hl=en_GB&pli=1#

and jose to draw workflow; then ALL of us to update the document, make sure it is clear what our problems and questions are and then organize telco with spyros and reto; or any of them;

• DONE: Danica to report to Ivan and Maurice the status of instructions from Sttudgart.

• * all-you-can-eat model, see this page

  • http://infochimps.com/ publish there?

  • list: get-theinfo@googlegroups.com

  • http://groups.google.com/group/get-theinfo

  • IN PROGRESS: make this page more appealing, those who see it need to understand what are the benefits for them; add examples; tasks for Danica (for LLD) and Jose to think about the same page about their Wikipedia dataset

  • DONE: Ivan: The latest version of ?FactForge is installed on /nfs_nat on .85 server so MPI

should be able to run experiments with it now.

Any Other Business?

n/a

Next telco

• End of May or June, as we have plenary in May