Extract from a Source Category

The Extract transform provides a one-step process to

  • use one or more properties of an existing category to define a new category with a defined relationship to the source category, and

  • extract the nodes and edges for the new category and relationship.

To illustrate, we’ll extract a new Author category and its nodes from a CSV table of bestselling books of 2025, available at Kaggle. Import by drag and drop gives us a single bestselling_Books_2025 category with a node created for each row in the table.

06 02 02 ExtractSource

From the source bestselling_books_2025, we’ll use Extract to define an Author category with the author’s name as a property, connected to the source through a new WRITTEN_BY relationship, and extract all the nodes and edges from the source.

To extract from a source category:

  1. De-select data on the canvas to extract from the entire data set.

    With nothing selected, transformations affect the whole graph. Otherwise, they operate only on the selection.

  2. Open the Transform panel and Extract subpanel.

  3. In Create Category:

    1. Select the source category from the dropdown menu. The menu includes all categories currently visible on the canvas.

    2. Enter a name for the new category (e.g. Author).

    3. Enter a name for the new relationship connecting the new category to the source (e.g. WRITTEN_BY).

  4. Select properties from the dropdown menu to be extracted from the source category as properties of the new category. Source property names are listed on the menu in alphabetical order.

    You can select as many properties from the source as you want. We only need one (e.g. Author).

    The selected property name appears in the New Property Name text box. You can enter a different property name; we’ll leave it unchanged.

    06 02 03 ExtractSetup

  5. Click the property to be used as the Key such that each unique value of the property creates a new node.

    1. Click Author, to create a separate node for each unique author.

    2. Click Split, because some bestsellers have more than one author, listed in the CSV Author column separated by commas.

    With Split selected, comma-separated values are extracted as separate nodes. Otherwise, a single node is created with only the first value in the list. Also,if the separator for multiple property values is not a comma, you can enter it in the text box.

  6. In Options:

    1. Click Skip Empty, since we want to prevent extraction of a node when the key value is empty, undefined, or null.

    2. Leave Inherit Relationships deselected. This is most often what you want when the source category has other defined relationships.

  7. Click the Run button at the bottom of the Extract panel. Log error and completion messages appear below the Run button.

    • The legend now includes the new Author category and the new WRITTEN_BY relationship.

    • Extracted AUTHOR nodes appear on the canvas, connected to their source bestselling_books_2025 nodes by the new WRITTEN_BY edges.

    06 02 04 ExtractCompleteLegend