Discovering related tables
You can engage the system to suggest the most appropriate tables for your data module. Using natural language processing and AI-based functionality, the system generates a data module that best represents your use case.
This functionality is used when creating a data module or adding new sources or tables to a data module.
The choice of tables for the data module is based on keywords that you select. An interactive word cloud visualization shows the keywords that exist in the available sources.
The following example shows a word cloud for a data module proposal that is based on four sources:
The font colors represent the different sources. The font size indicates the keyword weight, which is the measure of the keyword importance in the source. Selecting keywords with higher weight increases the probability of creating the most relevant data module for your use case.
To increase or decrease the number of keywords in the word cloud, expand the Keywords section in the right pane, and enter a number for the Keywords limit option.
You can select the keywords from the word cloud, or type them in the search bar. The selected keywords are automatically added to the search bar. To deselect the keywords, delete them from the search bar.
After you click Next, a data module proposal is generated, as shown in the following example:
The Proposed data module pane shows the tables that the system suggests to use for the data module. By default, one proposal is generated for each source. A percentage confidence score is assigned to each proposal. The confidence score reflects the predicted ability of the proposal to fulfill your modeling objective.
Select one or more of the suggested proposals. The selected proposals are merged into one proposal, and table relationships are generated.
Click OK to accept the proposal, or click Previous and try to generate different proposals.