Tech Stack
Same as version 1.3.1
New Features
API Endpoints
- FAQ RestHandler: New endpoint 'getFAQEntry ' created to return FAQ entries filtered by a field
Improvements
Server
- Server timeout is now configurable (it was hardcoded to 30 secs)
Recognizers/Processors
- FAQ: new setting to include or not the vector generated by the NN model in the response. Helps with minimizing the payload size when using the API.
UI/UX
- Minor updates on colors and labels
Aspire Saga-Parser
- Upgraded Aspire libraries to version 5.0.4
Export
- Export for Data Science was moved to "Evaluations" tab so it can be used with several tags at the same time. Once the export is finished, a link is made available for the user to download the file from the browser.
Documentation
- Documentation for version 1.3.1 is now up to date. Also added articles on how to migrate saga from environment to environment and how to configure Aspire-Saga-Parser in Aspire 4 and 5.
Bug fixes
- FAQ recognizer returned the vector fields repeated 3 times causing a gigant payload when using the API.
- Null pointer exception in Fragmentation stage when having several tags using the Fragmentation recognizer.
- Error when training a model in the Name recognizer.
- Entity recognizer not showing metadata fields in the processing results.
- Error in Entity recognizer when adding a new entry after an import.
- WebSockets not working when Saga is placed behind a proxy that has https set
- Sorting entries by column in the entity recognizer is not working
- Minor UI glitches
Note: you can find all the details in Jira here