# 3.4.4 Release Notes ## Features - LibNLP: Query Processing Pipeline: First iteration of semantic-term-expansion (dynamic synonym expansion) - Improved distribution of typeahead suggestions across all suggesters - Created plugin for project namespace in the configuration service - Removed the option from the studio plugin to specify whether project config values should be import/export. - Exposed project config interpolation - libNLP: Added Query-Processing example workflow to boost/filter terms within a user’s query - Added the ability to configure the facet containing the account name for the Insights Generator pipelet. - Enabled new project configuration - use Tika Server instead of inprocess Tika JVM - Handled project namespace for the configuration service in the squirro client - added config option which if enabled will load available react version of widgets - Handled new project configuration in the KEE plugin - Created project namespace for the configuration service - Applied user-scope for saved queries. Returns saved queries only from the user who saved them. - Changed Activity-Plugin’s incremental loading behaviour (state internally managed). This allows to use the plugin as a 1-click-connector and have incremental-loading working out-of-the-box, no manual configuration needed. - libNLP: Added Sentiment-Term-Normalizer; Extend VaderSentiment-Algorithm to extract positive&negative phrases from text (English only). - Removed the option from the squirro client to specify whether project config values should be import/export (obsolete because config-service’s project-namespace is used) - Updating the style of the dashboard edit - Deleted old project configuration plugin from the studio. - libNLP: Added SpacyNormalizer step to get cross-lingual POS, NER, Lemmatization, DependencyParsing etc. as an encapsulated step. Pipe text-content through configured spacy pipeline and forward the analysed spacy-document as an additional _nlp_ field. - Exposed project config interpolation in the squirro client - Initial version of Query Processing Pipeline based on ML-Workflows - 1st iteration of auto labeling process including inverse tf-idf stopwording, word2vec similarity + n-gram based topic generation - Moved popular query scope setting to the configuration service - Rerun from index now supports specifying a query (expressed in [Query Syntax](../../../../technical/search/features/query-syntax.md#search-query-syntax)). Only the items returned by this query will be rerun. - Replaced old project configuration with the new configuration service in datasource service and pipelet lib. - Disabled facet name suggestions when the query is empty - Handled importing and exporting for the new project config - added query field to rerun step modal. - Restore old order ## Bug Fixes - Fixed saved queries suggestions on the backend site - Fix PDF load race condition breaking the toolbar - Pass error of a request to the UI - Widget after removal seems to trigger change event and rerender the next widget 2 times - Fix scrollto in item detail - Add max height to chip - Fix the _–parallel-uploaders_ option of _squirro_data_load_ when used to upload items in row format. - ExtLinks are missing from PropertyInsightTypes - Fix gray color on edit buttons in Dashboard Loaders - Assure global_id is 22 characters length - Fix small overlay bug - Fixed invalid import path in the machine learning service tests - Fix a link color and a translation - Fixing text spreading out of the container - block negative proximity values for proximity filters in AI Studio - Fix sometimes-cant-scroll-to in pdfs - Modal step rerun improvements and bugfixes - In the cards component after loading it seems to load next items without the user scrolling - Tags editing style is broken and outdated - fixed container sizing issues for rules labeling - navigate back to prev page after saving dashboard instead of settings page - When link is _null_, the item is adding a link of the page/item - Improvement: Do not use nlp-tagger’s extracted topic-list for typeahead completion. - Do not group consecutive ML model steps into one when the _machinelearning.optimize-workflows_ option is set to false. - Pass error of a request to the UI ## Breaking Changes - Deleted old project configuration plugin from assets. It means that old plugin cannot be used any more. Instead, a new plugin using configuration service has been introduced. ## Installation and Upgrade You will have to resolve at least the following config files when upgrading from Squirro 3.3.0 - `/etc/nginx/conf.d/ssl.inc` - `/etc/squirro/common.ini` For new installations, please follow the [Installing Squirro on Linux](../../../install/linux.md#install-linux) instructions. To upgrade an existing installation, please consult [Upgrading Squirro](../../../install/upgrade.md#install-upgrade).