| Table of Contents |
|---|
Features
- Preprocessing capabilities to highlight entities of PDF documents when viewed in the PDF viewer.
New templates for AI Studio:
- Encode sentences with the pre-trained universal sentence encoder of google and use logistic regression to classify the sentences.
- Encode sentences with the pre-trained universal sentence encoder of google and use cosine similarity to classify the sentences. This template enables users to test their text classification use cases on a low effort base.
- On-premise users of squirro need to install the pre-trained model manually (`yum install squirro-google-universal-sentence-encoder`) before they can use the new templates.
...
Allow deleting data loader field templates
Bug Fixes
- Data source fixes:
Make the on-demand run of data sources to work for unscheduled (manual) sources (e.g. RSS feed without any defined scheduling options).
- Fix a bug where editing a data source could result in raising a security-related error.
- Facet fixes:
- Fix facet mapping issue in filesystem dataloader plugin on empty projects
- Fix Facet List widget not updating correctly after negative queries
- More lenient facet handling for Pipelets by allowing scalar facet-values like item[kw_float][facet_name] = 0.3. Priorly the ingester-pipeline accepted only values of type list like [0.3]
...
Fixing packaging of Transformers lib.
Known Issues
- Visualization / Custom Widgets / (add / edit): save button isn't visible but action works.
Configure Data Source: Enrich Pipeline section drop down overlaps with "Pipeline Workflow".
- AI Studio / Ground Truth / Focus view is broken. The Label view has to be used as workaround.
Installation and Upgrade
You will have to resolve at least the following config files when upgrading from Squirro 3.3.0
...