Exploring the Ontology
Overview
The ontology is a link between tables. You can relate the tables with a primary key (attribute in table 1) and foreign key (attribute in table 2) and can generate the Entity-Relationship Diagram (ERD) for these. The common identifier, a foreign key can be used to link the tables and generate relationships. You can utilize these relationships in the downstream process like Audience and Metrics.
There are three types of relationships:
- Non-editable system relationships: Created by the system as part of the data unification process, such as Stitch.
- Non-editable inherited relationships: Created automatically by importing data from data sources, such as Predict.
- Editable custom relationships: Created and configured by Skypoint AI users (API & Studio).
Skypoint AI lets you create two different types of relations (relational database designs, such as Entity-Relationship model (ER model)) between chosen tables.
Cardinality
The cardinality of a join between two tables is the numerical relationship between rows of one table and rows in the other. It is very useful for query optimization. They are of two types:
- One-to-One – A single row of the first table is associated with a single row of the second table. For example, if there are two tables, "customer" and "subscription plan". So, each customer can have only one subscription plan, and each subscription plan belongs to only one customer.
- Many-to-One – Many rows of the first table associated with a single row of the second table. For example, many orders created by a single customer. Therefore, there is a many-to-one relationship between orders and customer.
To create a Custom Relationship
- Go to Lakehouse > Ontology.
The Ontology window appears. The default view of the page displays all the relationships that have already been built. Each row represents a relationship.
Item | Description |
---|---|
New relationship | Allows you to add new relationship. |
Visualizer | Allows you to filter the view based on certain pre-defined criteria like table type, relationship types. You can see a network diagram of the existing relationships and their cardinality. |
Filter by | Option to filter the data based on relationships type (User, Inherited, System). |
Name | Indicates the purpose of the relationship. |
Source table | Indicates the table that is used as a source in the relationship. |
Source cardinality | Indicates the relationship between two relations in a data model. Cardinality describes the number of possible elements in a set. |
Target table | Indicates the table that is used as a target in the relationship. |
Target cardinality | Indicates the cardinality of the target table records. |
Type | Indicates the relationship type (User, System). |
Action | Allows you to edit or remove the relationship. |
Group by | View the items either in a list or a specific Group (For example, Name, Source table) |
- Select New relationship.
- In the New relationship pane, provide the following information:
To | Do |
---|---|
Enter the name that reflects the purpose of the relationship | Mention in the Name box. |
Describe the relationship | Explain in the Description box. |
Enter the table that is used as a source | In the Source details column, select the Table from the drop-down list. |
Enter the table that is used as a target | In the Target details column, select the Table from the drop-down list. |
Enter the cardinality of the source table | In the Source details column, select Cardinality from the drop-down list. |
Enter the cardinality of the target table | In the Target details column, select Cardinality from the drop-down list. |
Enter the foreign key field in the source | In the Equivalent attributes area, select the Source attribute from the drop-down list. |
Enter the primary key of the target table | In the Equivalent attributes area, select the Target attribute from the drop-down list. |
To enter the name of the relationship, start with a letter. Use letters and numbers only. Special characters and spaces are not allowed. Use between 3 and 64 characters.
- Click Save to apply your changes.
- To cancel your changes, click Cancel.
Using Search feature in drop-down lists
You can use the Search feature in drop-down lists to find your attributes quickly. This feature prevents unlimited scrolling while searching for correct attributes from the lists.
The Search feature is only disabled for the drop-down lists in Predictions.
Edit a Relationship
You can update properties of custom relationships that are already created.
- Navigate to Lakehouse > Ontology, and you will see all the created relationships in a tabular form with all their details.
- Look for the relationship that you want to modify. Click the horizontal ellipsis under Actions and select Edit.
- A new window will appear with all the ontology details. To apply changes, click Save.
Visualize Ontology
The visualizer navigates the ontology network to explore patterns of connection. It displays an ER diagram between connected tables and their cardinality. You can view how tables are connected. By default, the system is configured to a vertical layout.
- Click on the Visualizer to explore the relationship.
- Click Filter by to select different criteria such as user type, system type tables, or relationships.
- To view the ontology diagram, do the following:
To | Do |
---|---|
Maximize the display | Click the Maximize button. |
Minimize the display | Click the Minimize button. |
Restore size and display the complete diagram | Click the Fullscreen button. |
Drag and fix the display boxes | Click the Lock button. |
- You can change the layout of the tables and relationships from vertical to horizontal and vice-versa.
- To change the layout, click on the Change to horizontal or vertical layout.