A new minor release is available for Arkindex instances. You can test it on our demo instance: demo.arkindex.org
You can learn more about Arkindex on the official documentation
The most notable improvements for this release are visible while navigating in projects and their elements on an Arkindex instance.
The web interface will now build a unique URL for a specific element highlighted on a page, by simply clicking on a children element in the tree on the left side of the display.
Here is the unique URL to share that specific Act in the context of its parent Page.
You can now display more elements in a folder, using the Pagination size option in the Display menu. Possible values are:
- 20 elements (by default)
- 50 elements
- 100 elements
- 500 elements (this can be slow to load, depending on the filters used)
The selected page size is preserved when switching to the next or previous page.
Filter elements created by users
Finally, you can filter all the elements created by users (by opposition to Machine Learning workers). This is achieved by selecting the No worker version in the worker version filter while browsing elements. This can be useful to track manually annotated content.
As always, we aim to enhance performance on common API endpoints. For this release we focused on the ListElements endpoint, especially when using a
We also now use a cache for some part of the corpus retrieval on that endpoint, and simplified some SQL queries for the RetrieveCorpus endpoint.
You may notice the speed improvement with previous Arkindex releases, while browsing a large project on a specific type.
You can now import TIFF files directly from the upload page on a project, they will be converted to JPEG and immediately available on the Arkindex instance IIIF server.
Two new advanced settings are now available when starting a new process:
- an experimental feature to enable a local cache system between worker,
- another experimental feature to track workers activity on elements.
These two features are under development and will be documented in the upcoming releases.
- Support polygons up to 300 points without triggering API errors,
- Hide password reset form on successful reset,
- Exclude rejected classifications in the best class filter when filtering elements.
- The first and last name entities are not merged anymore as the current algorithm was producing invalid names,
- All API calls towards Arkindex are retried several times in the case of an API internal error (statuses 50x)