Building the Excel Workbook

To improve recognition accuracy, prepare the Excel file according to a consistent structure before importing it.

Recommended Structure

The recommended convention is:

One worksheet represents one data table.

Example workbook:

ecommerce_model.xlsx
├── Users
├── User_Address
├── Categories
├── Products
├── Orders
├── Order_Items
├── Payments
└── Logistics

The worksheet name is used as the default table name.

Column Headers

The first row of each worksheet should contain field names. Sample data should begin from the second row.

Users Worksheet

user_id	username	email	phone	level	created_at
1	Alice	alice@example.com	13800000001	VIP	2026-07-01
2	Bob	bob@example.com	13800000002	Normal	2026-07-02

Orders Worksheet

order_id	user_id	address_id	order_time	total_amount	status
10001	1	101	2026-07-10 10:30:00	299.00	paid
10002	2	102	2026-07-11 14:20:00	499.00	shipped

Table Naming Conventions

Use one consistent naming style, such as:

Users
User_Address
Order_Items

or:

users
user_address
order_items

Recommendations:

• Use English names
• Avoid spaces
• Avoid special characters
• Keep singular/plural conventions consistent
• Avoid mixing Chinese, English, and pinyin
• Use names that clearly describe the business entity

Not recommended:

Table1
Sheet1
User Table
order-items!

Field Naming Conventions

The lowercase snake_case convention is recommended:

user_id
product_name
order_time
total_amount
created_at

Recommendations:

• Name a primary key as singular_table_name_id
• Keep foreign-key names consistent with the referenced primary key
• Use _time, _date, or _at for date and time fields
• Use status consistently for status fields
• Use is_, has_, or enable_ prefixes for Boolean fields

Examples:

is_default
is_active
has_invoice
created_at
updated_at

Field Type Inference

The system infers field types from the content in Excel.

Excel Content	Recommended Inferred Type
1, 2, 3	integer
12.5, 299.00	decimal / float
User names, addresses, descriptions	string / text
2026-07-24	date
2026-07-24 10:30:00	datetime
TRUE / FALSE	boolean

To avoid incorrect inference, sample values within the same column should use a consistent data type.

Not recommended:

amount
100
200.5
N/A
Undetermined

Recommended:

amount
100.00
200.50
0.00
350.00

Primary-Key Design

Each table should have a unique primary key.

Examples:

Users.user_id
Products.product_id
Orders.order_id
Payments.payment_id

A primary key should be:

• Unique
• Non-null
• Type-stable
• Unlikely to change with business content

Names, phone numbers, and product names should not normally be used directly as primary keys.

Null Values and Sample Data

The system may use sample data to infer field types and relationships. Each worksheet should therefore contain approximately 2–10 valid sample rows.

Important guidelines:

• Do not merge cells
• Do not place description rows above the header
• Do not place multiple independent tables in one worksheet
• Do not use colors as the only representation of field meaning
• Do not insert summary or total rows in the data area
• Avoid completely empty columns
• Use a consistent date format

Multi-Worksheet Excel Example

Users
  user_id
  username
  email

Products
  product_id
  category_id
  product_name
  price

Orders
  order_id
  user_id
  order_time
  total_amount

Order_Items
  item_id
  order_id
  product_id
  quantity
  unit_price

After import, the system can generate a model containing relationships among users, products, orders, and order items.