How to Create a Knowledge Base: Basic Setup

Basic Setup

General Information

You can define the name of your knowledge base and add a description on the following page:

Retrieval Chunks

The number of retrieval chunks represents the maximum number of data chunks the AI assistant will reference when answering. The system default is "12", meaning the AI assistant will retrieve the 12 most relevant chunks for each response.

Therefore, you can increase or decrease the number of retrieval chunks to adjust the amount of information the AI assistant references when answering.

What is a Parser?

A Parser allows the system to "understand" the content within PDF documents, making it searchable, editable, or convertible to other formats.

When uploading PDF or Word format files, the parser provides three options:

MaiAgent Parser (default)
MaiAgent Parser (Online)
MaiAgent Parser (OCR beta)

The system currently uses MaiAgent Parser by default, which is particularly suitable for extracting structured data from complex professional documents for AI system analysis.

To understand the differences between the three parsers, please refer to: Technical Manual - Parser Tool

If problems occur during data parsing, you can also click the [Reparse] icon to have the parser reorganize the data.

Retrieval Model Settings

In the knowledge base settings, you can select the Embedding model and Reranker model you wish to use.

Embedding Model

Embedding is like translating human language into "numerical language" that AI can understand, allowing computers to comprehend the true meaning of text. This process is called "vectorization". Different Embedding models have different characteristics, such as languages they excel at processing and supported deployment environments. Different model settings within the knowledge base can be used to adjust the vectorization processing effect when knowledge base documents are uploaded. You can select the most suitable Embedding model for different scenarios.

You can freely choose from multiple Embedding models:

To understand the differences between Embedding models, please refer to: Technical Manual - Embedding Models

Reranker Model

Reranker is like a professional judge who re-evaluates which data in the initial search results can best answer the customer's question. What are the differences in effectiveness when using or not using a Reranker?

When a customer asks: "What tent is suitable for beginners? Budget under 8000 dollars"

Without Reranker:

AI might respond:
"We have tents at various price points, products at 8000 dollars include..."
(May mention advanced models, not focused enough on beginner needs)

With Reranker:

AI responds:
"Specifically recommended for beginners are these tents under 8000 dollars..."
(Precisely targets beginners + budget + product recommendations)

When search result reranking is enabled, the AI assistant will reorder the knowledge base content chunks retrieved, responding based on the most relevant documents.

To understand Reranker models, please refer to: Technical Manual - Reranker Models

In summary, using Embedding combined with Reranker allows the AI assistant to understand the knowledge you provide and, after retrieving chunks, review the importance of content again, responding with knowledge most relevant to the question.

Associate AI Assistants

Associating AI assistants means authorizing which AI assistants you want to use this knowledge base. If you have two AI assistants:

Product Customer Service AI
Order Customer Service AI

When both need to answer questions related to returns, you can associate both AI assistants at once in the settings of the "Return Policy" knowledge base:

Select the AI assistants to associate

Click Add AI Assistant

After adding, it will appear in the selected AI assistants area. Click "Save" in the lower right corner to complete the association.

Once associated, both AI assistants can share the "Return Policy" knowledge base and respond based on the same content. For subsequent maintenance, you only need to update one knowledge base to ensure the AI assistants use the latest data.