Version 3

20-06-2018

Version 3 of the Media Suite has new data, bug fixes and many new features!:

New data

ASR

A significant portion of the collections of The Netherlands Institute for Sound and Vision is automatically annotated using automatic speech recognition processes (ASR), resulting in verbatim transcripts available for search and interactive navigation of the resources. The automatic annotation process is still ongoing. In Version 3 transcripts are available for:

  • Radio 1 (Hilversum 1, NPO Radio 1): more than 90%

  • Radio 5 (747 AM, NPO Radio 5): more than 60%

  • Source catalogs (items from the Radio Programma, Weken Nederlandse Radio, and Hoorspelen collections): more than 40%

  • Television (news and current affairs): about 25%

EYE Jean Desmet

We added new collections and increased access to resources from collections from EYE Film Museum:

  • The Jean Desmet business (paper) archive has been made fully available for browsing (according to the same archival structure provided by EYE).

  • The Desmet business (paper) archive can also be searched, thanks to experiments with OCR-ing this collection. The OCR allows for searching the full text of the typed written documents. Other automatic detection experiments allow to group items per type (e.g., letters, telegrams, etc.), logos, signatures, languages, among others. See more information about how this collection was made available in the Media Suite data registration system. <>http://mediasuitedata.clariah.nl/dataset/desmet-paper-collection.

  • Viewing of the +100.000 digitized images is possible via a IIIF server.

  • Those images can be annotated.

  • The Desmet film collection, which metadata was made available in Version 2 of the Media Suite linking to the publicly accessible Youtube videos, is now fully available for viewing, more than eight hundred films can be played from the EYE servers via the Media Suite (available from July 2, 2018).

Oral history interviews (open access available)

  • The “open/open access” oral history interviews at DANS can be now played and annotated via the Media Suite.

Improvements to Beeld en Geluid Audio-visual collection

  • There were improvements to the data completeness of The Netherlands Institute of Sound and Vision after detailed data integrity checks.

KB newspapers

  • The index of the KB newspaper basic collection is fully completed (available for search from July 2, 2018)

Tools improvement

  • The Inspector tool has been redesigned and improved with “metadata dictionaries”.

  • We redesigned the Compare tool to work with the newly implemented Saved queries.

  • We fixed a number of bugs in the Search tool. We also added new features that were requested by the main users (researchers) to all tools. These include:

    • Improved faceted search (full term view, filtering, and download)

    • Saved queries

    • Improvements to the timeline charts and histograms

    • Improvements to the Bookmarking facilities

  • In the browsing tool (“Explore”) we made possible to connect the resources to the Workspace functionalities. Users can now add bookmarks and annotate resources from the Explore tool, save “Tool sessions,” and find all these in their Workspace.

  • The browsing tool (“Explore”), new functionalities were added to facilitate adding comments to “Browsing paths.”

Workspace (second version)

We improved significantly the “Workspace,” which first version was launched in the previous release of the Media Suite:

  • All views (bookmarks, codes, comments, links, metadata cards) were split for improving clarity

  • Each view was improved with filtering options

  • Viewing and exporting user fragments became more clear

  • Users can add personal collections to their work space, these collections can be searched and annotated (ASR for personal collections is forthcoming).

  • We added experimental Jupyter notebooks to allow python skillful users to work with the Media Suite data

And more…

  • We improved interface and interaction.

  • The Documentation pages and blog in Github pages have been fully updated.

  • Significant back-end improvements.