Page History

De Coronado, Sherri    

Safran, Tracy

Kuntipuram, Kumar

x

Current Sprint - Sprint 15 (October 15, 2015 - October 28, 2015)

LexEVS 6.4 S13-500 Agile Development - Sprint StatusStatus#SprintStatus-Sprint15

LEXEVS-1240 -  Searching for parent of a root noted returns the entity with code @@

• Kim discussed this issue and Tracy suggested this change would affect clients. Tracy will add her comments to the issue.

LEXEVS- 970 Is the overall OWL2 loader issue.

• TODO :NCI and Mayo to break this down to smaller issues to work on in the next sprint.

Issue Prioritization

Recent LexEVS Related Bugs and Features (within last week)

Recent CTS2 Service Related Issues (within last week)

Issue Grooming

Prioritize JIRA issues (fix version 6.4)

Property type investigation and Implementation

• Overview by Scott

Lucene doesn't have a "query" type mechanism.  When "AND"ing these properties together, the results are different than expected. We are looking into how to implement with the latest Lucene 5.3.1. 

TODO: Scott to create a diagram/description of the problem on the wiki (design documents section) to help everyone understand.

Sprint Deliverables - Updates

• 6.3.x
• Lucene

The build is currently broke.  The artifact built does not run due to the new lucene jars and their manifest files. 

The dev team will need to consider if we need to change the process to create a zip file, rather than a jar.

The dev team is suggesting breaking the build up into multiple jars.

This new multiple jar approach would break current users of the single jar.  Users would need to update their classpath.

TODO: Add two groups to work with on the NCI team to the Face to Face agenda.  How will this affect NCI users?

TODO: Mayo dev team to come up with a proposal on how to migrate to the new jar structure.

Sprint Deliverables

• 6.3.x
• Lucene

Mayo to create snapshots for both branches of code

6.3.0.1 - OBI fix and UI dropdown filter on coding schemes

Lucene - latest implementation

VPN and network evaluation

Craig

is still waiting to hear back from the Mayo

/NCI IT.

TODO: Craig will follow up with IT to get an update.

December 8 - 10 on-site.

There is a wiki link for agenda topics: LexEVS 6.3 and 6.4 Technical Face-To-Face Meetings

1) Backwards compatibility, if no extra information is included in the lexevs search call the resultset should be (nearly) the same as we currently get.

• To the best of our knowledge, this is true.

2) Expose the ability to do Boolean queries over different user-specified fields (e.g. presentation & definition) a la https://lucene.apache.org/core/2_9_4/queryparsersyntax.html#Fields.

• We will need to discuss what the requirements would be.  The boolean query is mainly at the API level of Lucene and isn't exposed.

TODO:  NCI team to add additional information to the topics for the Face to Face meetings.

Search queries - accessing the database opposed to the Lucene index only

• Update

LEXEVS- 970

• OWL2 Issue breakdown update

Scott was able to load the thesaurus locally without any errors.

TODO: Scott to work with Gilberto and review a load with restrictions.

Gilberto noted the following:

• He is not seeing any relationships for biotin (code CHEBI_15956) on his load.
• Role Relationships not displayed for code (OBI_0100061)
• has associated axiom (fol) (IAO_00100000) (the declaration of the property) -
  
  • Do we have this data in LexEVS db? 
  • Once we have this in the DB, can the TERM browser render this properly?
• information content entity - has curation status
• entity - editor note: annotation on annotation name is not rendered

The Mayo team will be out next Wednesday.

The NCI will cancel the meeting.

The Mayo team will end the existing sprint and start the next one

We are re-implementing how we are restricting coded nodes.

We are are seeing some differences in sorting when doing multi coding scheme index searches.

If you do a union query across multiple code systems.  Each code system has it's own index.  Scoring is done on a each index.  This is causing a junit ordering test to fail.

We will need to see how this works with real data.

If the CUI is left blank will it cause a problem?  We don't think so

Discuss how to determine whether a search query requires accessing the database opposed to the Lucene index only.

.