2.FAQ Knowledge Base Initialization Setup

After uploading and parsing the internal knowledge materials owned by the enterprise through the Agent Cloud backend, the Q&A format knowledge is stored in the knowledge base.

Once the knowledge creation is successful, agents can better learn the enterprise knowledge materials and respond to customer inquiries by associating with relevant knowledge points.

Based on the enterprise knowledge and system knowledge classification, knowledge sources can be divided into two types:

High-quality knowledge sources: Common FAQs accumulated by the enterprise are considered high-quality knowledge sources. They support bulk import using system templates. This type of FAQ-based knowledge is easier for large models to understand and apply, resulting in higher accuracy and faster response. Especially when the similarity between customer queries and knowledge points is 0.95 or above, the corresponding knowledge points will be directly matched for response, with the response time controlled within 1 second.
Example:
Question: How do I open the cap of the cleaning solution bottle?
Response: Hello, the cap of our cleaning solution is a safety cap. You need to press and rotate the cap to open it, in order to prevent children from accidentally consuming it.

Ordinary knowledge sources: PDF, Word, and other formats of enterprise product manuals, employee training documents, etc., are considered ordinary knowledge sources. After uploading, these knowledge sources will be parsed into knowledge fragments that are only split by the model and have not been manually processed.
For this type of knowledge source, the model can only extract text information and cannot extract content in the form of images, tables, etc. Therefore, there is a higher risk of answer errors due to missing knowledge.
Therefore, we highly recommend uploading high-quality knowledge sources to improve the accuracy of Q&A.
Therefore, we recommend uploading high-quality knowledge sources to improve the accuracy of Q&A.