IW Meeting 2013-03-14

From Inference Web

Jump to: navigation, search

IW Meeting 2013-03-14


Meeting Information


  • Administrative

- Patrick is creating a test version of IW Wiki that uses LDAP. He/We will let people know when it is ready to test. This *should* work the way it used to in the good ways and we should not drop functionality....


Patrice Jim Paulo Tim Deborah Cynthia Patrick

Meeting Preparation

Around the room

* Add a section for yourself 2 hours before meeting.
* Mark any discussion point that you would like to raise during meeting (with DURING MEETING). 
* Otherwise, assume that others will read the rest before meeting. 
* Also, please be considerate and read others' discussion points before the meeting starts.


First means Awesome. Proxy on VM issues. Waiting on Evan to look at it. LDAP (new) vs. the really really ^ 4 that's on onto. Version of media wiki.


Reviewed Paulo's PML 3 mapping spreadsheet comments with Deborah and Jim.



  • Received feedback on WebSci datacube/provenance paper
    • In general, reviewers liked ideas covered, but felt examples provided were weak. Reviewing the feedback now, and trying to work out an alternate submission.
  • Ongoing FUSE discussions. Developing Pivot demo, which will be ready by next week's Ontology/Explanation telecon.
  • from deborah- james - you and tim should coordinate so that james can see some of the work tim is doing that would support drill down encodings. Tim is working on a VAST paper that is related. (I told tim this in a meeting yesterday with him)


  • FUSE 2 demo: Publications, Patent Assignees, Patent Applicants.


PML3 paper writing: brief discussion about timeframe and format

Is my data set flagged (DQR)? Is there a new version of the data? Is there a new non-flagged (DQR) version of the data? Is there a data citation url for my data set embedded in my data that knows which copy of data I have and it's provenance? Is there a data citation reference for a set of data instead of just single files? Can I know what my algorithm parameters, traceable data input references, software stack and environment, traceable metadata input references for a particular data set? How can I compare the provenance of 2 iterations of data? What handle is used? when is it created? What resolution of time scale/data sets does it apply to? Can I compare the data from 2 iterations? How is an interation defined? Can an interation be 1 hour, 1 day, 1 week , 1 year? How is sub-settted variations communicated? If a week of data is affected within my year of data, how can my questions be answered? How are conflicting layers of meta data and data iterations reconciled to answer provenance and quality questions? How can we rectify that gap between data and dqrs? Are data sets defined by range, files, or common conditions (IE ordered the same day, use the same DOD,)? Can 2 users easily tell if they are using the same data? Or at least some portion of the same data?


  • Thanks Patrick for setting up my inference web account, it works.
  • Working on Annotator tool backend bits (project was have a DataONE internship proposal in for)
    • successful conversion from java, some bootscrap bits remaining
    • manipulating RDF model for adding enhancements.
    • working on bringing Robin along on frontend bits for passing csv file as parameter, and grabbing the column header id to enable updating enhancements file.


  • extended discussion with Tim and Jim about PML mapping
  • return to PML consensus pass
  • Patrick is moving the IW Wiki onto LDAP. He is creating a separate test version first to make sure things still work before the transition.



pml-j hasConclusion (Tim proposes DEPRECATED)

  • Paulo: How many conclusions can a plan derive? Aka? "how many conclusion information entities can a inferencestep activity derived"

Paulo: PML-J could only provide 1 conclusion (was considered a feature). NodeSets were a way to represent multiple conclusions with multiple node sets. NodeSet represented a conjunction of justifications. Each justificaion may have multiple antecedents.

2) source vs. sink are handled with just prov:wasDerivedFrom between pml:InformationContainers.

Paulo: in a way to have SourceUsage, we want a "SinkUsage".

A source usage is the derivation of an Information from an InformationContainer.

  a pml:InformationContainer;
  prov:wasDerivedFrom :works_for_accounting;
  prov:qualifiedDerivation  [ 
     a prov:Derivation;   # This is Paulo's "SinkUsage" - 
                      # the derivation of an InformationContainer from an Information.
     prov:entity :works_for_accounting;
     prov:hadActivity :the_insert;
     foo:bar "";

     a pml:Information;
     prov:value ":joe rdf:type Accounting";

  a prov:Activity;
prov:used :query_string;
prov:used :connection_server_access_thingy;

3) https://docs.google.com/spreadsheet/ccc?key=0ArTeDpS4-nUDdFdBYkRUQ0hvUWFpTy1ObjlZSTJEaXc#gid=0

spcdis pages on the web http://tw.rpi.edu/web/project/SPCDIS/usecases

Facts about IW Meeting 2013-03-14RDF feed
Date14 March 2013  +
Personal tools