InvoiceIQ 2.1 Release Enhancements

This release introduces a range of exciting and highly requested features aimed at enhancing the InvoiceIQ workflow for our clients:

• Added Recategorization block: A new Recategorization block has been added for recategorizing predictions, re-running MitraCode rules, and creating a re-categorization sync object. And developed logic for un-running and re-running mitracode rules for line items when re-categorization is applied to the line items.
• Implemented model retraining service: Developed a flowchart and gathered the requirements for creating a retraining service after researching and reviewing Quovant's retraining code. Developed a Docker environment for model retraining service and implemented categorization functionality.
• Improved InvoiceIQ connection settings: The new enhancement automatically starts the schedule on startup based on appropriate API key level. Clients can validate that their API keys are working efficiently by clicking the ‘Test Connection' button. And can trigger the prediction/training syncs immediately by clicking the ‘Sync Now’ , when Prediction Server and Training Server return a Connection Established status.

Here are the new and updated articles to deep dive into these changes:

• Module Installation and Configuration
• Recategorization

Enhancements/Improvements

Enhancement: Modifications in IIQ connection settings 
Description: IIQ Sync Actions are moved from scheduled actions to background quartz, this automatically starts the schedule on startup, based on appropriate API key level. Connection URL, Training connection URL and Training API key options are also wiped off from the IIQ Settings page. Public Read Only and Test will only run the prediction syncs. Whereas, Public Shared and Private will run both the prediction and training syncs. 
                            Clients can validate that their API keys are working and their instance can communicate with Mitratech's InvoiceIQ infrastructure by clicking the ‘Test Connection' button. If the Prediction Server and Training Server returned a Connection Established status, clients can trigger the prediction/training syncs immediately by clicking the ‘Sync Now’ button.
Tracking Code: TC-49520, TC-49521

Enhancement: Created softcoded url endpoints
Description: Public IIQ endpoints (pandora/chiron) endpoints are managed within the module as the libraries pull these values based on API keys; these values are editable by support if the endpoints change at any time.
Tracking Code: TC-51104

Enhancement: Modifications in categorization backend logic
Description: Backend logic has been modified to send incorrect categorization information to Chiron, so that the screen updates appropriately. Additionally, the backend logic was modified to reassign incorrect categorizations when triggered by line item users.
Tracking Code: TC-49511, TC-49512

Enhancement: Added Line Item block for Recategorization.
Description: A new interactive line items block has been added for recategorizing predictions, re-running MitraCode rules, and creating a re-categorization sync object. This recategorization objects are picked up by the Sync object. Developed logic for un-running and re-running mitracode rules for line items when re-categorization is applied to the line items.
Tracking Code: TC-49513, TC-49514

Enhancement: Updated IIQ settings for recategorization admin control option 
Description: A new checkbox 'Allow Recategorization reporting' has been added to the IIQ settings page to provide admin control over recategorization.
Tracking Code: TC-49515

Enhancement: Developed a flowchart and gathered the requirements for creating a retraining service after researching and reviewing Quovant's retraining code.
Tracking Code: TC-49531

Enhancement: Built docker environment for model retraining service and implemented categorization functionalities
Tracking Code: TC-50003, TC-50004

Enhancement: Built the secondary prediction endpoint along with Logic
Description: Created a secondary prediction endpoint in Pandora's Box for getting predictions from an intermediate model generated by the model retraining service, along with logic for generating predictions from the intermediate model. Also created new data anonymization service endpoints and retraining service API endpoints in Pandora's Box.
Tracking Code: TC-50005, TC-50006, TC-50007, TC-50008

Enhancement: Setup and configured Chiron DB and NFS
Description: Setup and configured a Network File Storage(NFS) and linked it to required docker services for storing all the model retraining data and files to be able access across Pandora’s Box. Tables are set up and configured in Chiron DB for storing model metadata which is accessed by the retraining service, which is configured to save and update model metadata information in Chiron.
Tracking Code: TC-50009, TC-50011

Enhancement: For training data selection, logic has been implemented to use invoice date instead of dataset name as a main filter. Modified the training database structure to make it more performance-efficient.
Tracking Code: TC-51101, C-51102

Bug Fixes

Issue:  Public Key Length validation has been added.
Tracking Code: TC-51508

Issue: Fixed the issue in finding line item data using ClientID.
Tracking Code: TC-50294

Issue: Fixed the issue with sanitization of dates.
Tracking Code: TC-50295

Issue: Fixed the issue of MRR$ Objects are not getting created after Recategorization.
Tracking Code: TC-50917

Issue: Removed duplicate ‘All InvoiceIQ Recategorize Request’ Tab under ‘InvoiceIQ Recategorize Requests’ tab.
Tracking Code: TC-50938

Issue: Fixed the issue of redirecting subscribe button of InvoiceIQ Recategorize Requests tab to Login Page
Tracking Code: TC-51116

Issue: Added license check to IIQ sync object rules to avoid creating objects, MRR$ and MLT$, for License Type ‘Public - Read Only’ and ‘Public - Test’.
Tracking Code: TC-51484