ISG Testing Cycle 2

From Islandora

Jump to: navigation, search

The following notes describe the status of the main features currently implemented on the Islandora Test Server,


Islandora 7 Test Server Features


Please review and comment on branding:

  • The entire site has been branded as a mock-up of branding for an individual site
  • The Basic Image collection has been branded
  • All individual objects are being branded with logos linked to "owner_inst" field.
    • preferred Logo files for branding are being collected.
  • Additional object branding, i.e. "other logo" not yet implemented.
  • Additional nodes/pages from site/collection not yet implemented. (under investigation)

Descriptive Metadata

The descriptive metadata entry and edit forms have been implemented:

  • MODS Simple Entry form implemented.
  • Full MODS edit form implemented.
  • See notes below on metadata transformations. The MODS to MARC and MARC to MODS transformations are expected to be a work in progress for some time.

MARCXML Collection records

  • Currently the FLVC edit forms do not support MARCXML <collection> records. Question: should we strip the <collection> level elements upon ingest? RESOLVED: Strip <collection> tags during ingest. 4/19/2013.

Notes on Metadata Transformations

The status of the transformations in Islandora are as follows:

  • Islandora is using the latest version of the MARC-to-MODS from the Library of Congress (revision 1.85 from March 7, 2013) with addition improvements from FLVC. While there will likely be hiccups that we notice while testing, for the most part this produces clean, decent and through MODS from any MARCXML file.
  • Islandora is the latest version of the MODS-to-MARC transform from the LOC (from May 2012), with some additional FLVC improvements. While this is the "latest" version from LOC, it's still a year behind the MARC-to-MODS and therefore doesn't produce MARC of the same quality as the MODS. Caitlin has written to the Library of Congress asking if they intend to update the XSLT any time soon. She will also post to the MODS listserv asking if anyone in the community has a newer / better version. In the meantime, we can work to make local improvements, but it's quite a large and complex document, so this will take some time.
  • Finally, we are using a highly-revised version of the LOC MARC-to-HTML xslt that is producing clean and nice HTML MARC from the MARCXML. The problem is that if there are any quirks in the HTML it's likely that the MARCXML underneath is flawed. We are working to solve those problems in various ways (see above).

User Permissions

Permissions work is not yet complete because Islandora v.7's XACML editor is not yet available. (The XACML Editor is both a GUI for editing XACML policies and a mechanism for enforcing all XACML policies.) FLVC will need to contract for this work with DGI if it's required for Day 1 of Production. Currently available permissions:

  • User Susan Submitter is supposed to only edit datastreams on objects owned by her (set by Drupal in the Owner field under the Manage -> Properties tab).
  • User Susan Submitter does not have access to the Properties tab and cannot change object Owner.
  • When Susan Submitter attempts to edit an object she does not own/has not submitted, an error appears on the screen. However, the edit is made. This will be corrected via XACML permissions. Currently the error message only alerts us that she shouldn't have permission to edit the object.
  • Permissions on all other Users are unchanged.

Indexing and Searching

  • The "simple search" (Search the Repository) includes all MODS elements but excludes full text.
    • Simple Search can be executed in a specific collection.
  • Advanced search currently includes the following indexes, and can easily be modified:
    • Title
    • Name
    • Full Text (from PDFs, either loaded with PDF or generated by Islandora during ingest if no text file is provided).
    • Submitting Institution
    • Owning Institution
    • Date Created
    • Date Issued
    • A pull-down to limit the Advanced Search by Collection. Default is All Collections. (The Search button needs to be moved.)
  • A "view all objects" button has been implemented, which essentially just does an "empty simple search".
  • Note:
    • Please let FLVC know if you've ingested or modified a record and don't see the indexing change reflected in search.
    • Solr is calling solr.EnglishPorterFilterFactory to do stemming when indexing text fields, and also when querying text fields - "reduces any of the forms of a word such as "walks, walking, walked", to its elemental root e.g., "walk" ". We can remove this call for queries. Or it can be removed from both indexing and queries.
    • The highlighted snippets feature has been turned on in the results list for searched of "Full Text" from Advanced Search.
    • Because "view all objects" uses simple search, we had changed simple search to exclude page objects. But simple search was still searching the full text of PDF objects. It seemed confusing for it to be doing a full text search for PDF objects but not books. I changed the indexing, so now simple search only searches non-fulltext fields (dc, mods, etc) and excludes page objects. If you decide that simple search should include full text fields and page objects, we'll have to come up with a different approach for the "view all objects" button.
    • MODS field indexing is a work in progress. Fields with type="date" are indexing as Solr type string and text for display and searching. If it has a recognizable date format, it will also index as a Solr date field. Current MODS field names do not include the type in the name. The newest MODS configuration attempts to do this, but in an overly broad way. We will have to decide how to proceed.
  • Facets have been implemented:
    • Only relevant facets will display in search results.
    • Current facets (easily adjusted):
      • Collection
      • Subject Topic
      • Subject Geographic
      • Genre
      • Type of Resource
      • Language
      • Owning Institution

Books Module

  • A single book was loaded into the Books collection/Content Model using the default Islandora Books Module to demonstrate the IA Book Reader in Islandora.
  • The custom FLVC code to convert METS to JSON format was not used in this instance and METS was not used in any way for this book example.
  • Discussions are underway regarding the use of the JSON format for generating a structured TOC file and display.
Personal tools