From Filtered Push Wiki
Jump to: navigation, search

Etherpad for compiling meeting notes: http://firuta.huh.harvard.edu:9000/FP-2011Feb01


  • Paul
    • Anne Marie reports that Dell has gotten back to her about disks for firuta, but with more questions.
    • Revised sequence diagrams for alternative view of proposed architecture. Pulled together on a List_of_Component_Responsibilities
    • Sent out a textural description of the database mapping problem to Bertram, Lei, and Zhimin, with context and examples of mapping issues with dates and new identifications.
  • Lei and Bertram
    • Read paper about Lens “Relational Lenses: A Language for Updatable Views” and tried its prototype. Currently the prototype seems mostly for text mapping while not provide support for relational database mapping.
    • Discussed with Zhimin on requirement of data schema mapping in FPush project and mapping example to start.
  • Zhimin
  • Bob
    • More non-success with SMW+. May start a clean installation if it.
    • Started separating AO extensions from instance examples. OWL API does not like instances without class declarations.


  1. Potential Collaboration with GeoLocate
  2. CodeHosting
  3. Schema Mapping
  4. Architecture Progress
  5. Harvard ODT report
  6. Second Project Programmer Position

Meeting Notes

Filtered Push meeting notes, 2011 Feb 1.

Present: Zhimin, Jim, David, Paul, Chinua, Bertram, Bob, James.


  • Potential Collaboration with GeoLocate

Potential for working with geolocate to georefence specimen data associated with the gulf coast. http://www.museum.tulane.edu/geolocate/ Existing network, movement into the vertebrate collection networks.

Short list of possible (most probable) code forges. We should move off the UMB infrastructure to something better able to handle the project management. Prime contenders: source forge, google code. Untill recnetly google code hasn't had good connections between issue tracking and code repository (coments on changes to code can embed standard references to bugs, and this can automatically show up and influence the state of the bug in the issue tracker (avoiding thus both commiting code and commenting on bug status)). Prinicple issue with google now is the proprietary issue of the issue tracker - making migration from google code difficult. Issue with a subset of international contributors, probably applies to both google code and source forge, but more generally to google code.

Bob moved Specify6 from the Kansas repository to Sourceforge, straightforward and easeir if done early.

Plan: Paul to start sourceforge project, bob to migrate svn data in two weeks for now.

  • Schema Mapping

Zhimin: Examining Arctos mapping to flat and bulkloader.

6 classes of mapping

Field to field

Format conversion (e.g. date format conversion)

Two fields on one side to one field on the other (e.g. day/month/year)

Data conversion of one field, but dependent on another. Depth field, unit field, convert to global schema using depth in meters, need to understand unit to convert.

Fields from several records into one field. Separate table for collector, collecting event has several collectors. Global schema has single concept for collector. Retrieve collector rows, concatenate into one field.

Inverse of the previous, global schema has list of multiple concepts, local has one.

Bertram: Is going to pass on use case example. Looking into clio. Orchestra, news is mixed. Is an academic prototype, worked in principle, but not a production system. Clio might be more productized.

Paul: also constraints.

Bertram: Yes, mappings are essentially queries themselves and embed constraints.

Bob: A problem of inverting a view?

Bertram: At least in part. Question, do we just want to do the inverted mapping, materialized view approach, or virtual integration through a mediator, mediator rewrites the query on the fly.

  • Architecture Progress




Bob: Where things get done probably get tokens (e.g. at query initiation time), token gets passed around, services check token, rather than continuous authorization. When we get to sequence diagrams, authorization will happen where it needs to, pieces in the middle probably need to delegate some authorization to some peers (node 1 asks question it is entitled to the answer, but client is not, node 2 responds with some data that node one can formulate an answer that the client is authorized for). Users of nodes may not be the only players getting authorized. There is likely to be internode authorization. We don't want to preclude complex webs of authorization resulting from a query that doesn't need much authorization.

Zhimin: Authetication is a separate service from both nodes and users.

Bob: Yes, both authentication and authorization separate, and authorization to use services looks potentially separate from authorization to query data.

  • Harvard ODT report

In progress.

  • Second Project Programmer Position
Time to compile short list.