Knowledge Base Tutorial

In version 0.9.1, CherryStudio introduced the long-awaited knowledge base feature.

Below we will present detailed instructions for using CherryStudio step by step.

Add embedding model

1. In the model management service, find models — you can click "Embedding Models" to quickly filter;

2. Find the desired model and add it to My Models.

Create a knowledge base

1. Knowledge base entry: In CherryStudio's left toolbar, click the knowledge base icon to enter the management page;

2. Add knowledge base: Click Add to start creating a knowledge base;

3. Name: Enter the name of the knowledge base and add an embedding model; using bge-m3 as an example, the creation is complete.

Add files and vectorize

1. Add files: Click the Add File button to open the file selector;

2. Select files: Choose supported file formats such as pdf, docx, pptx, xlsx, txt, md, mdx, etc., and open them;

3. Vectorization: The system will automatically perform vectorization. When it shows complete (green ✓), it means vectorization is finished.

Add data from multiple sources

CherryStudio supports multiple ways to add data:

1. Folder directory: You can add an entire folder directory; supported files in that directory will be automatically vectorized;

2. Website URL: Supports website URLs, such ashttps://docs.siliconflow.cn/introduction;

3. Sitemap: Supports sitemap files in XML format, such ashttps://docs.siliconflow.cn/sitemap.xml;

4. Plain text notes: Supports entering custom plain-text content.

Tip:

1. Illustrations in documents imported into the knowledge base are not currently supported for conversion to vectors and need to be manually converted to text;

2. Using a website as a knowledge base source may not always succeed — some sites have strict anti-scraping mechanisms (or require login, authorization, etc.), so this method may not retrieve accurate content. It is recommended to run a test search after creation.

3. Most websites provide a sitemap, such as CherryStudio'ssitemap, which can usually be found by adding /sitemap.xml to the site's root address (the URL). For example,aaa.com/sitemap.xml .

4. If a site does not provide a sitemap or the URLs are messy, you can assemble an XML sitemap file yourself; the file currently needs to be provided as a publicly accessible direct link — local file links will not be recognized.

1. You can have the AI generate a sitemap file or have the AI write an HTML sitemap generator tool;

2. Direct links can be generated using OSS direct links or cloud drive direct links. If you don't have ready tools, you can also go toocoolAIofficial website and, after logging in, use the site's top-bar free file upload tool to generate a direct link.

Search the knowledge base

After files and other materials are vectorized, you can perform queries:

1. Click the Search Knowledge Base button at the bottom of the page;

2. Enter the query content;

3. Display the search results;

4. And show the matching score for that result.

Generate replies in conversation by referencing the knowledge base

1. Create a new topic, then in the conversation toolbar click Knowledge Base to expand the list of created knowledge bases and choose the one to reference;

2. Enter and send a question; the model will return an answer generated from retrieval results;

3. At the same time, the referenced data sources will be attached below the answer for quick access to the source files.