Datasets

A Dataset defines what data to extract from a web page and how to structure it.

Dataset Structure

Each dataset has:

• Name — A unique identifier for the dataset
• Item Selector — The CSS/XPath selector for the repeating parent container (e.g., .product-card)
• Properties — Individual fields to extract (e.g., title, price, image)
• Required Fields — Fields that must have a value for the record to be valid
• Parent Dataset — Optional link to a parent dataset for nested data

Creating a Dataset

Via AI Assistant

Ask the AI to extract specific data. It will create the dataset definition automatically.

Via Visual Builder

1. Open the project in the Visual Builder.
2. Switch to the Data Extraction tab.
3. Click + to add a new dataset.
4. Click + to add properties.
5. Use the element picker (pointer icon) to select elements on the page.

Via AI Visual Extraction

1. In the Data Extraction tab, click the wand icon (✨).
2. RTILA takes a full-page screenshot and sends it to the AI.
3. The AI identifies data structures and creates datasets automatically.

Item Selector

The Item Selector identifies the repeating container element. For example, on a product listing page:

Item Selector: css=.product-card

This tells RTILA: “Find all elements matching .product-card. Each one is a separate record.”

All property selectors inside the dataset are relative to the item selector.

Properties

Each property extracts one field from each item:

Field	Description
Name	The column name in the output (e.g., “title”, “price”)
Selector	CSS, XPath, Text, or Variable selector relative to the item
Type	What to extract: text, html, attribute, property, count, page_url, selector_path, list, index
Attribute	(For attribute type) Which attribute to read (e.g., href, src)
Property	(For property type) Which DOM property to read
Download	If checked and the value is a URL, the file is downloaded automatically
Transformations	A chain of transformations to apply to the extracted value

Dataset Previews

The Data Extraction tab shows a live preview table of the data that would be extracted.

Tip: If the preview shows “No matching data found,” check your selectors. You can use the element picker to fix them.

Required Fields

Mark fields as required in the dataset settings. If a required field is empty for a record, a warning is shown in the preview.

Nested Datasets (List Properties)

You can create list-type properties that extract nested data. For example, extracting all reviews for each product:

1. Add a property with type list.
2. Set its selector to the review container (e.g., .review).
3. Add sub-properties inside the list property (e.g., reviewer name, rating, comment).