By integrating with platforms such as Ollama, WeKnora can flexibly switch between mainstream big models such as Qwen and DeepSeek. The system first converts private documents into vectors and stores them in the local database. When a user asks a question, it first retrieves the relevant document fragment, and then inputs the fragment and the question into the big model to generate an answer. This design not only utilizes the reasoning ability of the big model, but also ensures that the answer is based on the enterprise's private data, avoiding the factual errors of the general model. The whole process data is completed in the enterprise intranet, forming a secure closed loop.
This answer comes from the articleWeKnora: Tencent's out-of-the-box enterprise-level Q&A knowledge baseThe
English 
简体中文 
日本語 
Deutsch 
Português do Brasil