Click the Basic Configuration button in the robot card zone, where you can peek at or edit the essential robot information, welcoming phrases, message responses, and Q&A strategies.
Basic information
Click Edit on the Basic Information card to enter the editing status.
In the editing mode, you can modify the robot's avatar and name. Upon completion, click Save.
Note:
After the robot name and avatar are successfully saved, they will take effect in the current network, please operate with caution.
Welcome message Setting
Welcome message script: Upon entering the conversation, the robot will send the first welcome message.
Welcome Card: Upon entering the conversation, the bot will automatically push the problem configured here in card form, which can be used to display high-frequency common questions.
You can configure the title and questions for the welcome card, and it supports up to 20 questions.
You can sort the questions on the welcome card by using the sorting button on the left side of the question, the robot will by default display the questions that are sorted higher.
Response message effect
Two response effects are supported:
Streaming output: The robot's reply content presents a streaming output animation effect. After the feature is enabled, if you need to achieve a streaming effect, please refer to the Chatbot Integration Guide to complete the development. Synchronous output: The robot's reply returns directly. For Q&A policies like the affinity large language model, the return time might be long, thus we suggest selecting the streaming output.
Setting for Q&A Policy
We provide you with the following Q&A policies, where Direct response is the default policy. In addition, you can choose to enable the Casual conversation strategy and the Guided questioning based on your business situation.
|
Direct answer | The bot directly addresses the user's queries. When the similarity between a user's question and the question in question bank reaches the established direct answer threshold, the bot will provide the answer corresponding to the query with the highest similarity. |
Small talk | The bot responds to casual salutations by users. Upon enabling the feature, if user's inquiry fails to reach the direct answer threshold set in the Q&A library, yet similarities to greeting library reach the set small talk threshold, the bot will respond directly with the answer corresponding to the highest similar greeting. |
Clarification | The bot counter-questions the user's intended queries. After enabling the feature, when a user's question does not meet the direct answer threshold set by the Q&A library, and does not reach the small talk threshold , but it exceeds the clarification threshold , the robot will return the questions with the highest similarity. |
Fallback Response Rule | When user's question is not recognized by the Q&A library or the small talk library, the robot triggers a fallback response, supporting the following two fallback patterns: Fixed fallback: Upon triggering the fallback response, the chatbot will respond with the fallback dialogue configured here. Large language model fallback: After building an affinity with the large language model engine , it can be enabled. When a fallback response is triggered, it will invoke the large language model engine affiliated with you. |
Note:
Threshold settings will impact the performance of the robot's responses. Unless there are special requirements, the default threshold is recommended.
Was this page helpful?