November / December 2022
about 2 years ago by Hicham TAHIRI
2.23.0
- Smartly.AI :
18th November
- Virtual Agent :
7th December
Autocompletion
- This new feature allows you to help users with suggested queries (sourced form the bot intents) as they type their query in the chat. This feature is currently available via the web chat or API.
You can try it in the live demo below:
- New: Make the autocompletion robust to case variations and diacritics
- New: Make the autocompletion work for masterbots
- New: Make the autocompletion work after a jump or a language switch
- New: Make the autocompletion available via API
- New: In the Analytics module, distinguish between messages suggested by the autocompletion from the others
API
- Improved: New
input_type
value namedautocomple
for the/dialog
endpoint - New:
/autocomple
endpoint is now available.
Webchat
- Improved: Faster response time for the first message of a bot answer
- Improved: New
reduce
andrestart
icons
Other features
- My Bots: Ability to search for a bot by its ID
- Builder: In the available Node.js environment, the
google-auth-library
has been updated to the 8.6.0 version - Security: Screen lock time has been increased from 10 min to 30 min
- Inbox: Better pagination management while working on the supervised learning
2.23.1
- Smartly.AI :
28th November
- Virtual Agent :
7th December
Autocompletion
- Improved: Adaptative scroll bar
- Impreved: Better support for RTL languages
2.23.2
- Smartly.AI :
22th December
- Virtual Agent :
7th December
In this version we shipped many important updates to the NLU models
SNP 1.05 RC4
- Correct entities formatting
- Correct logging error
- Correct decimal separator for better number entity detection
- Correct training error related to multi-threading
- Correct API's URL prefix
- Fix seed for random sample of synonyms
- Fix augmentation seed
- Fix Augly version
- Fix training seed for reproducibility
- Don't preprocess entities
- Add logs to train service
- Improve environment variable management
- Correct formatting error to improve custom entities detection
SNP 1.06 RC11
- Remove pre-processing from the exact match data
- Set a seed for the data augmentation process (same seed as 105)
- Set a seed for the CRF model (used for entity extraction)
- Fixed an error in the logs that caused the train to fail
- Add more logs on the train
- Fixed entity extraction on SNP 1.05 and SNP 1.06 for the exact match
- Fixed extraction of entities with special characters
- Fixed problem of entity detection when lemmatization is activated
- Improved NLU pre-processing order (better understanding and extraction of entities)