đĒFlag and Train
Introducing Orimon AI's Flagging and Training Feature - A Simple Solution for Continuous Learning
Last updated
Introducing Orimon AI's Flagging and Training Feature - A Simple Solution for Continuous Learning
Last updated
Orimon AI platform offers an intuitive and powerful solution for continuous learning. One of its key features is the Flagging Feature that enables users to flag incorrect responses during test chats or from chat history. The flagged responses are then marked and made available for simple correction in the training section. This feature helps to enhance the chat-bot's knowledge by enabling trainers to update the correct responses and train the bot for better performance.
To use this feature, follow these simple steps:
During test chats invoked either from testing page on Orimon Dashboard or using test bot links, if you or your team comes across any response that doesn't seem right and needs to be corrected, click on the "Edit & Train" button visible on the respective response message.
Following this, a prompt will open-up asking you to make the necessary modifications to the Chat-bot's output.
Once you have made the necessary changes, you can either "Save" this change for training later or click on "Save & Train" to train the changes now.
From chat history, select the message that needs to be corrected and click on the "Flag" button.
Following this, a prompt will open-up asking you to make the necessary modifications to the Chat-bot's output.
Once you have made the necessary changes, you can either "Save" this change for training.
All the query-response pairs which are flagged in previous steps will be visible in the training history section and collated together for correction.
Go to the "Test & Train" page once again -> the click on "History" at the top of your chat-bot.
A dialog box will then open up as follows.
Once all the flagged messages have been corrected, train the bot with the updated responses.Once training is successfully completed you will be redirected to testing screen where you can verify your bot with the updated knowledge you have trained with.
Navigate to Chat History -> Bot Training Opportunities
This section serves as a focal point for refining your bot's intelligence. Here, you'll find interaction pairs flagged for review either by user feedback or our AI observability feature.
How It Works:
User Message: The user's initial message that triggered the bot's response. Bot Response: The bot's reply that has been flagged for review.
Feedback Source: Indicates whether the feedback was provided by a user or identified through AI analysis.
Issue Identified: A brief description of the potential issue with the bot's response. Action Taken: Current status of the interaction, such as updated, pending, or dismissed.
Train/Ignore: Options for how to proceed with the flagged interaction, suggesting whether to refine the bot's response or not prioritise it. Use these insights to continuously enhance your bot's accuracy and relevance. Each adjustment is a step towards a more intelligent and reliable chat-bot for your users.
