Changes between Initial Version and Version 1 of Taskforces/Curation/Meeting20160511

Timestamp:

06/13/16 09:23:58 (8 years ago)

Author:

davor.ostojic@oeaw.ac.at

Comment:

--

Taskforces/Curation/Meeting20160511

@@ +1 @@
+[[PageOutline]]
+
+= Curation TF - Centre Meeting 10-12 May, 2016 =
+{{{
+#!html
+ <table style="border-collapse: collapse;" border="1">
+  <tr>
+    <td>Date</td>
+    <td>11. 05. 2016</td>
+  </tr>
+  <tr>
+    <td>Time</td>
+    <td>15:10 - 17:00</td>
+  </tr>
+  <tr>
+    <td>Location</td>
+    <td>SURF Offices, Utrecht, NL</td>
+  </tr>
+  <tr>
+    <td rowspan="8">Participants</td>
+    <td>Claus Zinn</td>
+  </tr>
+  <tr><td>Davor Ostojic</td></tr>
+        <tr><td>Francesca Frontini</td></tr>
+        <tr><td>Henk van den Heuvel</td></tr>
+        <tr><td>Menzo Windhouwer</td></tr>
+        <tr><td>Mitchell Seaton</td></tr>
+        <tr><td>Neeme Kahusk</td></tr>
+        <tr><td>Pavel Stranak</td></tr>
+</table> 
+}}}
+
+# Agenda #
+
+* Curation Module
+* Availability / License Facet
+* Normalisation Workflow
+* !ResourceClass Facet
+
+
+# Notes #
+
+Remark: This document gives an overview over discussed topics during the meeting.
+Order of the topics in this document may not correspond to the original order from the meeting!
+Some additional resources might be added.
+
+== [wiki:"Curation Module"] ==
+
+Menzo: add to the collection view a column for the profiles with scores and link profiles to the view page. It will be helpful for the Meerents institute to identify critical profiles
+
+Pavel: OLAC has similar tool and result representation is very good. It offers hints how to improve the score. 
+
+Davor: there is a plan to develop such an application ([https://moqups.com/ostojic.davor@gmail.com/L9GiyDLr mockups])
+
+Davor: For profiles assessment both xsd and xml are needed. What happens when schemalocation attribute points the different location then component registry? In that case only xsd is available but for fetching xml url can reconstructed with profileID from CMDI’s header. Which schema to use then, from component registry or originally specified? (example: http://hdl.handle.net/11858/00-203Z-0000-0027-536B-3) 
+
+Menzo: this situation should be considered as invalid. Schemas could be out of sync. Give penalty to the score for this. Schema from component registry should be used. 
+
+Davor: There are situations when <!MdProfile> element has an url (component registry by the way) instead of profileId (clarin.eu:cr1:p_...). Hard to handle all of them
+
+Menzo: There are no strict rules what is allowed for <!MdProfile> element but id is preferred. It should be reported by the tool. Schematron could be use for such a validation. 
+
+Davor: Should Schematron be consider in curation module?
+
+Menzo: why not it is an ISO standard. [[browser:CMDIValidator|CMDI Validator]] from Oliver Schonefeld is using it already. Validation rules can be externalized.
+
+Davor: Score calculation still needs to be coded (maybe this can be externalised as well)
+
+
+== [wiki:Taskforces/Curation/ValueNormalization/License "Availability / License Facet"] ==
+
+Participants are informed about what has been changed:
+* [https://docs.google.com/document/d/13__Kde_iOxvZsYpKZkmFuint5whzPda806VENFvxHUc/edit?usp=sharing google doc]
+* [wiki:Taskforces/Curation/Meeting20151015 CAC2015]
+
+
+Pavel: it is very confusing which values to use for availability / license. Would be nice to have guidelines
+
+Claus: CLARIN provides licence category calculator
+
+Pavel reported problems for LINDAT’s data regarding this facet. Needs to be investigated whether VLO importer interprets original values in a wrong way or necessary fields are not specified by LINDAT. It seems that curation instance of VLO (with old mappings) gives correct result.
+
+Agreement is to try to fix LINDATs data together and then from gathered experience to try to create some guidelines or best practices for others.
+
+
+Email from Florian Schiel sent on 29/04/2016:
+
+''The module report a missing facet 'rightsHolder' on instances of profile 'media-corpus-profile', but the profile (and the instances I tested) contains the element <Owner> which is linked to concept ​http://hdl.handle.net/11459/CCR_C-2956_519a4aab-2f76-0fd3-090e-f0d6b81a7dbb''
+
+
+Email from Hanna Hedeland sent on 02/05/2016:
+
+''Since I put this facet in the facetConcepts.xml sometime during the VLO-TF time, I was quite sure we should have this information, but in fact it's now represented by http://hdl.handle.net/11459/CCR_C-2956_519a4aab-2f76-0fd3-090e-f0d6b81a7dbb in our profiles, and this concept is not in facetsConcepts. On the other hand, I didn't know that there was a decision to use this facet (listed among the test facets), but I won't tell anyone if we just include it ;) So, if it should be regarded, could you please add this forth concept to the list in facetsConcepts or discuss it at the next meeting or however things are handled right now?''
+
+__Ticket is issued: #931__
+
+Email from Hanna Hedeland sent on 09/05/2016:
+
+''I'm afraid at least some of the issues I described after the launch of the latest VLO release are still there, I think they could be defined as mapping problems?
+Even if http://hdl.handle.net/11459/CCR_C-5439_98bb103d-476a-7f62-54b4-bf9de24d2229 is provided, there are still entries with double values for distribution type (which should not be possible?), e.g. for https://vlo.clarin.eu/record?q=hzsk&docId=http_58__47__47_hdl.handle.net_47_11022_47_0000-0000-534B-F.
+It is not clear how to achieve that VLO entries get the laundry tag "PLAN", and entries with seemingly identical CMDI "legally relevant" metadata don't seem to receive the identical set of laundry tags (https://vlo.clarin.eu/record?q=hzsk&docId=http_58__47__47_hdl.handle.net_47_11022_47_0000-0000-631F-F and https://vlo.clarin.eu/record?q=hzsk&docId=http_58__47__47_hdl.handle.net_47_11022_47_0000-0000-5C5F-0).''
+
+__Menzo: we need to check mappings and importer.__
+ 
+
+Pavel: superprofile should be consider, a profile containing mandatory fields like author, title, license, resourceType, from which all others are derived.
+
+== Normalisation workflow ==
+
+
+Davor / Henk: How to discover values that haven’t been normalised?
+
+Menzo: Developers can query SOLR and obtain values.
+
+Davor: collect them during the import process.
+
+Davor: How to maintain controlled vocabularies, updating them and ensure that VLO always uses the latest version?
+
+Menzo: github might be a proper solution. It gives nice preview for csv files.  Push unmapped values to the github file (maintainer will be notified automatically).
+
+Francesca: Github provides users with in-browser editor
+
+== [wiki:Taskforces/Curation/ValueNormalization/ResourceType ResourceType] ==
+
+
+Davor: Normalisation map created by Jan Odijk is still not used. Should we use it?
+
+No agreement on this (again). 
+
+Davor: how about to create a concept and to link it with controlled vocabulary and recommend it for resType
+
+Menzo: CCR TF was thinking about doing something similar for all facets. Another option is to create a component. 
+
+Henk: too many values without a hierarchy are senseless but creating a hierarchy could possibly lead to display issues  (maybe they can’t fit the screen). 
+
+# Documents #
+
+* [[https://docs.google.com/document/d/1u0bS_PNP6R9dySHSoPHxyepNBPIwOqZJbb9bW3qkX9U/edit#|gdoc notes]]
+