Features

• Expose Newsletter improvements (expose domain to Newsletter templates; Add , add unsubscribe link to my account; Remove , remove capitalization from item titles; Remove , remove exclamation mark from newsletter subject)

• Improve user guidance in the ai studio AI Studio .

• Improve create community screen.

• Splitting of Content Standardization step into 3 steps ( Sanitize HTML, Remove HTML and Content Standardization ). Where Sanitize HTML cleans html document of potential malicious html tags, Remove HTML removes html from fields and Content Standardization makes sure that the item as the correct structure to get indexed.

• Disable the "Create ground-truth" button if a dataset is not created

• Improve html tokenizer to also tokenize titles

• Added GroundTruth ID as a parameter of machinelearning_template request

• Add 'sales_insight' application in the list of app-templates to be built.

• We now add the user preferences to Redis as soon as the plugin is loaded and the preferences endpoint is called for the upgrade cluster pluginWe have a new ´Sales Insight´ application template which can be used to investigate sales cloud data and its sentiment using a pre-trained ML models.

• Improve user guidance in the ai studio AI Studio

• Configuration service for easy configuration of advanced settings through the user interface. See Configuration service for details.

• Add smartfilter Smartfilter highlighting for GroundTruth labeling.

• When creating a new ground truth, sentence-level tagging will be preselected by default.

• Pre-select data loader recommended facet regardless of facet existing in the project.

• Added academy links to multiple pages and context sidebar.

• Add a new step called "Change Pipeline", to redirect the data to a different Pipeline

• Improved performance/throughput of the Pipelet-Step through parallelised mini-batching

• NLP-Tagger Pipelet improvements. See NLP Tagger

  • Allow usage of not installed / custom SpaCy language models (language code mapping)

  • Performance improvement for big PDFs (approximation): process only a subset of pages (configurable)

Bug Fixes

• Do not raw load the widget html. Fixes all custom widgets.

• Fix the communities documentation url.

• Fixed UI issues inside page context sidebar.

• Translations fixes.

• Changed the password in the smoke test to comply with the new password policy.

• Added missing variables in the webpack.

• Fix position of search icon in communities list page.

• Fixes z-index in the Timeline Widget dropdown.

• Fixed a bug with module not find for widgetsSamples and optimized webpack build performance.

• Don't cleanse trailing parentheses from facet values unless they start with parentheses too.

• Redis-server depends on policycoreutils-python-utils for Centos 8 installations.

• Wait for Es to finish indexing.

• Updated query regex to remove ")" from the string. Sort select dropdown should work as expected now.

• Import ml workflow independent on model.

• Workaround for the ssl keyword argument error that was raised during connection to machinelearning and plumber databases during the execution of the migrate_pipeline_steps.py migration script.

• Fix authentication loop which could occur during configuration of a OAuth2-enabled Data Connector.

• Fix redirect url upon creation of empty project.

• Improve entity highlighting for more complex html documents.

Known Issues

• Don’t upgrade to this release if you have custom pipelets in your Pipeline Workflow and your Project has some non-string Facets (datetime, int, float)

Installation and Upgrading

...

To upgrade an existing installation, please consult the Upgrades for Squirro 3.2.0 and later guide.