Make Data Available for Oracle Analytics AI Assistant and Home Page Ask

About Indexing a Dataset for Oracle Analytics AI Assistant and Home Page Ask

You must index a dataset so that its data is included in Oracle Analytics AI Assistant and Home page Ask.

Note:

Oracle Analytics AI Assistant is being rolled out in phases, based on region and shape. To check if the Assistant is available in your environment, display the Search tab of a dataset and confirm that you can index for the Assistant. See Index a Dataset.

If you don't see the Assistant option, you might not have the required permission, Use Assistant in Workbooks. Contact your administrator to request the permission. See Grant or Revoke Permissions for Application Roles.

You can index any dataset that you have Full Control or Read-Write access to. When you enable a dataset to be indexed, you'll specify if you want to index all or some of a dataset's attributes by names, or by names and values, and which language to index the dataset in. You'll also select whether the indexing is available to Home page Ask, Oracle Analytics AI Assistant, or both.

Note:

A file-based dataset must be indexed before you can use it to build visualizations from Homepage Ask. A file-based dataset must be indexed and certified before the users you shared the dataset with can use it to build visualizations from Home page Ask. See Generate Visualizations from the Home Page with Questions and Certify a Dataset.

File-based dataset indexes are refreshed when manually you reload the data, or on a schedule you determine. See Index a Dataset on Demand.

By default, the dataset is indexed whenever its data is refreshed so that the most current data is available when users search from the Home page. Your data is automatically refreshed any time a user saves changes to the dataset. Alternatively, you can add a schedule to indicate when and how often a dataset is indexed, or you can index a dataset on demand.

You use the Search tab on the Inspect page of your dataset to enable indexing based on files or data sources. Administrators can use the Data Model pane on the Console's Search Index page to control which subject areas are indexed. See Configure Search Indexing.

You can improve results from your Oracle Analytics AI Assistant with careful indexing and organization of your data. LLMs are better with text than numbers, so change null values to text, like ‘Unknown’ or ‘Not available’. You can also optimize your results by binning numerical values, like binning customers into groups based on total sales, or binning order amounts into ranges of small, medium, and large. See Create a Bin Column When You Prepare Data.

Enriching your data can provide greater context to the Assistant, making it easier to pull in data. For example, enriching countries with their continents means better results when prompting for “total sales in asia”. See Enrich and Transform Your Data.

You can reduce ambiguity in your indexed data by having clear column names that can be understood by anyone reading them, avoiding duplicate names, and leveraging synonyms. For example, datasets often have multiple date fields (order date, Ship date, invoice date). This can cause confusion when someone asks a question like “show sales by month”. To avoid this, create a synonym called ‘Date’ on the column you want to assign as the default date. See Specify Synonyms for Dataset Columns.

You can also improve results by carefully selecting which columns to index. Consider omitting any columns from the index that users are unlikely to prompt the Assistant for, like Row ID or Product Key. It’s also a good idea to avoid indexing columns that have a high cardinality, such as customer IDs, SKUs, or addresses.

WARNING:

The Oracle Analytics AI Assistant uses a Large Language Model (LLM) when processing your questions. As a result, it may generate responses based on patterns in data rather than factual accuracy. Always verify the results and consult your primary data sources before making critical decisions based on results generated by the Assistant.

About Recommended Index Settings

Recommended index settings are applied automatically to Search indexes after you select an Index Dataset For option.

Index settings are automatically determined based on cardinality and the data types of attributes in the dataset. Once the recommended index settings are applied, you can modify them individually. After you've changed any attribute settings, you can click the Use Recommended Index Settings button at any time to restore the system recommended index settings.

Index a Dataset

You must index a dataset to make its data attributes available to Home page Ask and Oracle Analytics AI Assistant. You can index all or some of the dataset's attributes and apply synonyms to make the attributes easier to search.

You can create a schedule to specify when and how often you want a dataset to be indexed. See Schedule When to Index a Dataset.

On the Home page or Data page, select a dataset.
Hover over a dataset, then click Actions, then Inspect.
Click the Search tab.
Click the Index Dataset For dropdown and select an option.
Click Index, select an option for indexing the data elements.
Select Use Recommended Index Settings to let the system determine your index settings.
Click the Languages field and select the language you want to use to produce the dataset's index.
Click Save.
Click Run Now to index your dataset immediately.
Click Refresh

to check the completion status of the index.

Index a Dataset on Demand

You don't have to wait for a dataset to be indexed after refresh, or wait for a dataset's indexing schedule to run. You can index a dataset anytime you need to make its data available in the Home Page's search results.

You can index any dataset that you have Full Control or Read-Write access to.

For information about enabling and setting up a dataset for indexing, see Index a Dataset.

On the Home page, click Navigator, and then click Data.
Click the Datasets tab.
Locate the dataset that you want to index on demand, click Actions, and then click Inspect.
Click the Search tab.

Description of the illustration dataset_search_index_run_now.png
Click Run Now to index the dataset.

Schedule When to Index a Dataset

You can schedule when and how often a dataset is indexed. You can choose to index the dataset at a time that is convenient for and doesn't interrupt or confuse workbook authors or consumers.

By default when indexing is enabled for a dataset, the dataset is indexed when it's refreshed. Setting an indexing schedule overrides this default. You can enable indexing for any datasets that you have Full Control or Read-Write access to.

For information about enabling and setting up a dataset for indexing, see Index a Dataset.

On the Home page, click Navigator, and then click Data.
Click the Datasets tab.
Locate the dataset that you want to add an indexing schedule to, click Actions, and then click Inspect.
Click the Search tab.
In the Indexing Schedule section, click Start and then click the Select Date Time button.

Description of the illustration dataset_search_index_schedule.png
In the dropdown calendar, browse to and select a month and day. At the bottom of the dropdown calendar, click the time stamp and specify what time you want indexing to run.
Go to the Repeat every field, enter a number. Click the dropdown and select how often you want the schedule to run.
Click Save.