How to Quickly Build a DeepSeek‑Powered Knowledge Base on Tencent Cloud

Introduction

The author recently completed a technical year‑end summary and began testing an internal knowledge‑base built on DeepSeek, now deployed on Tencent Cloud. Tencent Cloud has launched a full‑feature DeepSeek V3+R1 model with a large‑model knowledge engine and internet‑search capability.

1. DeepSeek Online Assistant

By clicking “立即体验” (Start Experience), the DeepSeek assistant on Tencent Cloud responds faster than the public version and automatically enables internet search, providing up‑to‑date answers.

2. Rapidly Building an Intelligent Knowledge‑Base Application

The process starts on the application creation homepage. Users can create an app, name it, and proceed to the operation console.

2.1 Application Configuration

Select the newly released DeepSeek‑R1 model, then fill in role instructions and a welcome message on the left panel. The “一键优化” (One‑Click Optimize) button refines the role prompt. Enable the “联网搜索” (Internet Search) option for real‑time web results.

2.2 Knowledge Management

Navigate to Knowledge Management and click Import . Choose “Import from Web” and paste the URL of the open‑source project documentation. The platform’s built‑in crawler fetches the pages, parses the content, and imports it into the document library. After import, test a question to see the retrieved document snippet and source link.

2.3 Knowledge‑Base Parameter Tuning

Adjust retrieval settings such as the number of top‑matching document fragments to return and the similarity threshold. Higher thresholds yield more precise answers.

2.4 Publishing Management

Once the app is ready, click Publish , fill a brief release note, and submit. Publishing completes within seconds, providing three sharing options: direct web experience, a public link, or a QR code.

3. Summary

The entire workflow takes less than ten minutes and requires no programming skills, making it accessible to all users. For developers needing deeper customization, the platform also offers an API, which will be covered in a future article.

For the full source and further tutorials, see the GitHub repository https://github.com/Java-Interview-Tutorial.