Helvia Release 5.28.0
08 March 2023
08 March 2023
We have made some updates and added more features to various pages to provide better UX and make your chatbot management even more handy!
The organization name and the organization avatar are now located to the top right cornet of the platform. Clicking on the organization name allows you to switch organizations if you are part of more than one orgs.
A filter has been added in the Bot's Activities page, enabling you to filter by activity type.
The bot Configuration page has a new UX, grouping settings together and making it easier to configure.
The Chat Sessions page has been redesigned and it has been enriched with new features. A new action button has been added next to each session, to show more information about the session.
Once the button is clicked, a window is expanded, with all the chat information, including the full conversation, the session ID, created dat, last interaction date, total messages, sentiment analysis, session duration, user name, user language, deployment name and origin, i.e. where the bot is hosted. It also gives you the option to add tags to be able to classify the conversations.
The date picker has been redesigned and implemented in all pages where a date selection is available.
Chat sessions will be exportable in an excel format (.xlsx). To export chat sessions in .xlsx, select 'Records' from a bot's page left hands side menu and on the 'Chat Sessions' tab click the 'Download' button. From the drop-down menu, select XLSX to download the sessions.
Admins have the option to deactivate a bot within the platform. To deactivate a bot, go to the bot's Settings page in the 'Configuration' tab, switch the 'Enable Bot' toggle in the Danger Zone section and click 'Save'.
If the bot is deactivated, it will not handle incoming messages. You can enable it again by turning the switch on.
When hovering over an Automated Answer, the ID will be displayed.
A new feature has become available allowing testing NLU models with test-sets. To test an NLU model of a bot with Automated Answers, select the 'Test NLU models' option from the left handside menu.
In the 'Bot Test NLU models' page you can manually add the test set or import from a spreadsheet that will include the User Input and the Expected Intent ID. Once the test-set has been added, click 'Run' to sun the test.
The results will include the below:
Detailed test results showing Macro Avg. Precision, Macro Avg. Recall, Macro Avg. F-1 Score, Accuracy
Classification report
Precision: number of correct predictions for an intent divided by the total number of predictions for the intent
Recall: number of correct predictions for an intent divided by the total number of expectations for the intent
F1 Score: a machine learning evaluation metric that measures a model's accuracy by combining the precision and recall scores of a model
NLU model performance analysis, showing User Input, Expected Intent, Detected Intent
Confusion matrix: overview of the predicted intent vs the expected intent
NLU models can be retrained through the bot's Settings page, in the 'NLU Models' tab. In this page click the 'Train NLU model' button to retrain an NLU model from the list.
LiveChat agents will be able to see the full conversation the end-user had with the chatbot, in the LiveChat platform. To view the conversation, click on the 'Show chat history' button in the conversation:
You will then see the full conversation the end-user had with the chatbot, before joining live-chat. You can scroll up and down to see the full history.
The 'End live-chat conversation' but has been redesigned for better UX.
In the cases where LiveChat is connected to Microsoft Dynamics, there is the option to have a link to the current contact in the CRM for optimal case management by the agent.