Manage How Content Is Indexed and Searched

As an administrator, you can index data sources and catalog content that enables users to find the latest content when using the search bar on the home page.

Topics

• About Search Indexing

• About Indexing Subject Areas for Oracle Analytics AI Assistant and Home Page Search

• About Adding Synonyms to Your Data Models

• About Managing Data Model Indexing and Synonyms Using a CSV File

• Tips on Using Synonyms for Home Page and AI Assistant Search

• Configure Data Model Search Indexing

• Configure Subject Area Search Indexing for Oracle Analytics AI Assistant

• Configure Subject Area Search Indexing for Home Page Search Visualizations

• Specify Synonyms for Data Model Columns from the Console

• Export and Import a CSV File Containing Synonyms for Data Model Columns

• Configure Catalog Search Indexing

• Schedule Regular Content Crawls

• Monitor Search Crawl Jobs

• Rerun a Search Crawl Job

• Certify a Dataset to Enable Users to Search It from the Home Page

About Search Indexing

Administrators can control which content is indexed and searchable on the Search Index page.

• Data model - Use the Data Model tab to configure which columns to include in the search index. See Configure Data Model Search Indexing.

  You can also control whether Oracle Analytics AI Assistant and home page searches can use the indexed data in their search results. See About Indexing Subject Areas for Oracle Analytics AI Assistant and Home Page Search.

• Catalog content - Use the Catalog tab to configure which objects to include in the search index. See Configure Catalog Search Indexing.

Note:

File-based datasets are indexed differently. Users who upload datasets decide how and when they're indexed through the dataset's Inspect dialog. See Make a Dataset's Data Available for Search.

Crawl results are added to the index in the languages that you specify. For example, if your company's headquarters are in the United States, and you have offices in Italy, then you can choose English and italiano to create an index in both English and Italian.

Data Model

When you select to index items in the data model, you can specify how the content is crawled using one of the following options for Crawl Status:

• Index Metadata Only: Indexes only dimension and measure names. This is the default selection. For example, column names such as Product or Order, and metric names such as # of Orders. Always use this option if the column contains sensitive data values that you don’t want to expose to users when they search on the home page.

• Index: Indexes metadata (dimension names and measure names) and data values. Applies only to dimension or attribute columns. For example, if you select this on a Product column, both the metadata for the Product column and its data values (such as iPad, iPod, iPhone ) are indexed.

  Indexing data values provides additional functionality for users who want to visualize data values from the search bar on the home page. Be aware that selecting this option can be costly because it indexes values for all of the columns in all subject areas of the semantic model.

  Note:

  When you index data, it's visible to all users who have access to that column. Take care not to index data for columns that contain sensitive data, as this will expose the sensitive data values on the home page.

• Don’t Index: Use this selection to completely exclude subject areas, tables, or columns from the index.

Catalog Content

When you select to index objects in the catalog, you can specify how the content is crawled using one of the following options for Crawl Status:

• Index: Use this option to include folders and items in the index.

• Don’t Index: Use this selection to exclude folders and items from the index.

  Oracle recommends that you don't set the Crawl Status field to Don't Index as a way of hiding an item from users. Users won't see the item in search results or on the home page, but are still able to access the item. Instead, use permissions to apply the proper security to the item.

You should select only the items needed to create useful search results. Indexing all items yields too many similar search results.

About Indexing Subject Areas for Oracle Analytics AI Assistant and Home Page Search

Administrators can control which subject areas in the data model are indexed and whether the Oracle Analytics AI Assistant and home page search can use the indexed data in their responses.

In the Data Model tab you can use the Assistant and Homepage Search checkboxes to determine if one or both can use the data in datasets or columns as part of their output.

• Assistant - When you enable the Assistant to use subject area content, you allow the associated data to be used in responses to natural language prompts.

• Homepage Search - When you enable Homepage Search to use subject area content, you allow the associate data to be used when generating visualizations with prompts from the home page search bar.

When configuring the data model index, you should be deliberate in choosing which data is indexed. If you choose to index all the data for the Assistant and Search, you might achieve poor prompt results and indexing more than 500 columns can lead to slow responses, reduced relevance, and increased processing costs. See Improve Responses from Oracle Analytics AI Assistant.

Note:

Avoid indexing data for columns that contain sensitive data, as this exposes those sensitive data values in any places where those columns can be searched and in responses from the Assistant.

The way you configure the index for datasets is different. See About Indexing a Dataset for Oracle Analytics AI Assistant and Home Page Ask.

About Adding Synonyms to Your Data Models

Administrators and business users can assign synonyms to data model columns to make it easier for users to find columns when they search for content from the home page or improve the responses fromOracle Analytics AI Assistant.

The names of data model columns are often times ambiguous to visualization authors and difficult for users to find when they use the home page search or AI Assistant. Oracle recommends that you use synonyms to clarify the meaning of your data model's column names to optimize search results and improve responses.

For example, to make it easier for users to find data in a column named Yield, you might add several synonyms such as revenue and income. This means users find data associated with the Yield column when they enter the terms revenue or income.

You can add synonyms to your data model in different ways:

• Administrators can add synonyms to columns from the Search Index page Data Model tab. See Specify Synonyms for Data Model Columns from the Console.
• Business users can use their favorite CSV editor to add synonyms to a CSV file exported by the administrator. After adding synonyms to the CSV file, the administrator imports the file to Oracle Analytics. See About Managing Data Model Indexing and Synonyms Using a CSV File and Export and Import a CSV File Containing Synonyms for Data Model Columns.

Data model synonyms are included when you take a snapshot of your environment. This enables you to preserve synonyms when you migrate between environments. Synonyms are also preserved when you edit the corresponding semantic model and redeploy your updates.

About Managing Data Model Indexing and Synonyms Using a CSV File

Administrators can manage subject area indexing and synonyms for AI Assistant and home page search using the Search Index page or they can export the data to a CSV file.

Administrators can use the CSV file to:

• Migrate AI-ready subject area metadata between environments (using the import CSV file option).

• Engage other business users to contribute to the data model metadata. Administrators can share the entire CSV file or segments of the file to subject matter experts and ask them to populate the CSV file with synonyms. Administrators can then import the completed file and apply the provided synonyms to enhance data understanding and insights.

Below is what you'll see if you open the CSV file with Excel. Note that the EnableAssistant, IndexType, EnableHomePageAsk, and Synonyms columns are the only columns business users should update. All changes in the CSV file are imported into Oracle Analytics, so be very careful to only change values in the columns highlighted below. Changes made to the other columns in the CSV file can introduce errors into the data model and break visualizations.

Description of the illustration csv_example_synonyms.png

About Specifying Indexing Properties and Synonyms

You can specify these indexing properties and synonyms in the CSV file:

• EnableAssistant - This cell is set to Y if the Assistant checkbox in the Data Model tab in the Console is selected. It is blank if the object or column's Assistant checkbox in the Data Model tab in the Console is not selected. In the CSV file, you can set this column to Y or leave it blank (no value).

  When you set this column to Y, you must also set any of the column's child elements to Y.

  When you set this column to Y, you can set the IndexType column to either IndexMetadataOnly or Index.

  If you set this column to N, the IndexType field must be blank.

  When you set this column to Y you can add synonyms to the Synonyms column.

• IndexType - This cell is set to Index Metadata Only or Index or is blank depending upon how the object or column's Crawl Status field in the Data Model tab in the Console is set.

  Note: the EnableAssistant cell needs to be set to Y for you to specify Index Metadata Only or Index.

  • Index Metadata Only - Indexes only dimension and measure names. For example, column names such as Product or Order, and metric names such as # of Orders. Always use this option if the column contains sensitive data values that you don’t want to expose to users when they search on the home page.
  • Index - Indexes metadata (dimension names and measure names) and data values. Applies only to dimension or attribute columns. For example, if you select this on a Product column, both the metadata for the Product column and its data values (such as iPad, iPod, iPhone ) are indexed.
  • No value - Excludes subject areas, tables, or columns from the index.

• EnableHomePageAsk - Indicates if want search to have access to the subject area data for visualizations. This cell is set to Y or is blank depending upon how the data model root level is set in the Data Model tab in the Console is set. If the data model root level is selected in the Data Model tab, this cell is set to Y for all child rows. If the data model root level is not selected in the Data Model tab, this cell is blank for all rows.

• Synonyms - Confirm that the corresponding EnableAssistant cell is set to Y and enter synonyms, using a comma to separate synonyms. Note: this cell contains any synonyms defined in the Synonyms field in the Data Model tab in the Console. See About Adding Synonyms to Your Data Models and Tips on Using Synonyms for Home Page and AI Assistant Search.

Tips on Using Synonyms for Home Page and AI Assistant Search

Use the following tips to help you create and assign data column synonyms.

If you own or have read-write access to a dataset, you can assign synonyms to its columns from the dataset's Search settings. See Specify Synonyms for Dataset Columns.

If you're an Administrator, you can assign synonyms to a data model from the Console's Search Index specifications. See Specify Synonyms.

If you're a business user, you can use the CSV file provided by an Administrator to assign synonyms to a data model. After you've finished assigning synonyms, the Administrator imports the CSV file into Oracle Analytics. See Specify Synonyms in an Exported CSV File.

Tips on specifying synonyms for column names:

• Enter one or multiple synonyms. For example, for a Yield column you might specify revenue and income.
• When updating a CSV file to add multiple synonyms to a data model column, use a comma to separate the synonyms. For example, revenue,income.
• Synonyms can be up to 50 characters.
• You can specify a maximum of 20 synonyms for each column name.
• You can use upper or lowercase when you specify synonyms. The home page and Oracle Analytics AI Assistant search isn't case sensitive.

You can't base synonyms on:

• Analytical functions, for example, sum, AND, OR, NOT, BETWEEN, IN, IS NULL, LIKE, Aggregate At, Aggregate By.
• Analytical terms, for example, null.
• Articles, prepositions, pronouns, and conjunctions that are usually removed before processing natural language (also known as stop words). For example, a, an, and, are, as, at, be, but, by.
• Boolean function names, for example, true, false, yes, no.
• Date formats, for example, nn/nnnn, nnnn/nn, nn/nn/nnnn, nnnn/nn/nn where n is an integer.
• Integers, for example, 123 or 123 456.
• Special characters, for example, `!@#$%^&*()+=[]{};\':"\\|,<>/?~.

Configure Data Model Search Indexing

You can configure how and when subject areas are indexed from the Data Model tab.

1. In the Oracle Analytics home page, click Navigator , then click Console.
2. In the Console under Configuration and Settings, click Search Index.
3. In the Data Model tab, click Enable Data Model Crawl.
4. In User to Run Crawl As, click Search and enter a user that is a member of the BIDataModelAuthor or BIServiceAdministrator application roles.
5. Select a language or hold down Ctrl to select multiple languages.
6. Under Select Data Models to Index select a Crawl Status for your data models and columns.
7. Enter Synonyms at the column level for use by your organization.
8. Click Save .
9. Optional: Click Run Now to rebuild the search index with your changes.

Configure Subject Area Search Indexing for Oracle Analytics AI Assistant

You can configure which indexed subject areas are available to the Oracle Analytics AI Assistant from the Data Model tab.

1. In the Oracle Analytics home page, click Navigator , then click Console.
2. In the Console under Configuration and Settings, click Search Index.
3. In the Data Model tab, click the Assistant checkbox for the subject area data you want the Assistant to have access to. You can control the specific columns Assistant can use data from by expanding the option and clicking the Assistant checkbox for only those columns.
4. Click Save.
5. Optional: Click Run Now to rebuild the search index with your changes.

Configure Subject Area Search Indexing for Home Page Search Visualizations

You can configure which indexed subject areas are available for creating visualizations from home page Search from the Data Model tab.

1. In the Oracle Analytics home page, click Navigator , then click Console.
2. In the Console under Configuration and Settings, click Search Index.
3. In the Data Model tab, click the Homepage Search checkbox for the subject area data you want the Search to have access to for visualizations. You can control the specific columns Search can use data from for visualizations by expanding the option and clicking the Homepage Search checkbox for only those columns.
4. Click Save .
5. Optional: Click Run Now to rebuild the search index with your changes.

Specify Synonyms for Data Model Columns from the Console

Administrators can use the Search Index page (Data Model tab) to add synonyms for specific data model columns to make them easier to search and find from the home page or Oracle Analytics AI Assistant.

To learn about how to assign data column synonyms, see Tips on Using Synonyms for Home Page and AI Assistant Search.

Administrators can perform this method of specifying data model synonyms from the Data Model tab on the Search Index page.

Note: Alternatively, administrators can export a CSV file containing the data model columns and distribute it to business users to update with synonyms. When completed, the administrator imports the modified CSV file containing synonyms into Oracle Analytics. See About Managing Data Model Indexing and Synonyms Using a CSV File.

1. In the Oracle Analytics home page, click Navigator , then click Console.
2. In the Console under Configuration and Settings, click Search Index.
3. In the Data Model tab, expand the data model layers to locate the level columns.
4. In the Synonyms column, locate the column you want to add a synonym to, click the synonyms field, and enter the term for the column and press Enter. Enter more synonyms as needed for the column.
5. Click Save, and then click Run Now to rebuild the search index with your changes.

Export and Import a CSV File Containing Synonyms for Data Model Columns

Business users can add data model column synonyms to a CSV file exported from Oracle Analytics by the administrator. After the administrator uploads the modified file, the synonyms are applied to the data model and make it easier for users to search and find columns from the home page or Oracle Analytics AI Assistant.

You can do partial imports of the CSV file. For example, instead of circulating one copy of the CSV file among several business users to provide synonyms, you can distribute copies of the CSV file that contain subsets of columns to several business users to gather their updates. You can then import each of the CSV file copies as users complete them. Remember that if you use this method and business users mistakenly provide synonyms for the same column, that the last import overwrites synonym changes for the same column from an earlier import.

Indexing starts automatically after you upload the file, so there is no need for you to run indexing manually. The index process is fast and the synonyms are implemented for use in the home page and AI Assistant in a few minutes.

Share these links with business users to help them properly assign data column synonyms: About Managing Data Model Indexing and Synonyms Using a CSV File and Tips on Using Synonyms for Home Page and AI Assistant Search.

Alternatively, the administrator can add synonyms to specific data model columns listed in the Data Model tab to make them easier to search and find from the home page or Oracle Analytics AI Assistant. See Specify Synonyms for Data Model Columns from the Console.

Tip:

You can also use the CSV import and export process to easily migrate AI-ready data model metadata between environments.

1. In the Oracle Analytics home page, click Navigator , then click Console.
2. In the Console under Configuration and Settings, click Search Index.
3. To export the data model as a CSV file:
   1. In the Data Model tab, click the data model to export as a CSV file. Clicking the model activates the Import and Export buttons.
   2. Click Export to create the CSV file and download it to your computer.
4. To import the updated CSV file containing new or updated synonyms:
   1. In the Data Model tab, click the data model to import as a CSV file. Clicking the model activates the Import and Export buttons.
   2. Click Import and in the Upload Crawl Config dialog, browse for and select the CSV file to import. Click OK.

Configure Catalog Search Indexing

You can configure how and when catalog objects are indexed from the Catalog tab.

1. In the Oracle Analytics home page, click Navigator , then click Console.
2. In the Console under Configuration and Settings, click Search Index.
3. In the Catalog tab, confirm that Index User Folders is selected.
   Oracle recommends that you don't deselect this option. When deselected, no folders in the catalog are indexed so home page searches return very limited and possibly no results.
4. Select a language or hold down Ctrl to select multiple languages.
5. Use the Catalog Object (Shared Folders) and Crawl Status columns to browse for and specify which catalog objects you want to index.
6. Click Save .
7. Optional: Click Run Now to rebuild the search index with your changes.

Schedule Regular Content Crawls

The administrator selects which folders to crawl and schedules when and how often to crawl the content.

1. In the Oracle Analytics home page, click Navigator , then click Console.
2. In the Console under Configuration and Settings, click Search Index.
3. Select Data Model or Catalog.
4. Use the Schedule options to specify when and how often to run the crawl.
   The index updates automatically as users add or modify content in the catalog.
   • Catalog crawl frequency: By default, a catalog crawl runs once a month. The minimum number of days you can specify between catalog crawls is 7 days.
   • Data model crawl frequency: By default, a data model (that is, semantic model) crawl runs once a day.

   In some cases you might want to schedule a crawl as needed (for example, after importing a BAR file or if automatic indexing didn't run).

5. Select a language or hold down Ctrl to select multiple languages.
   Crawl results are added to the index in the languages you specify.
6. Click to save your changes.

Monitor Search Crawl Jobs

Administrators can check the last time content was indexed and monitor the status of crawl jobs. You can stop any crawl job that is running, cancel the next scheduled crawl before it starts, or rerun a failed crawl.

The Crawl Job Status page shows information about the past, current, and the next scheduled crawl. In the Progress column, XSA indicates a dataset. If users report search issues, check the status of crawls to ensure that they're current. After a crawl is completed, users might have to wait a few minutes before they can locate the latest content.

1. In the Oracle Analytics home page, click Navigator , then click Console.
2. In the Console under Configuration and Settings, click Search Index.
3. Click Monitor Crawls.
4. Look at the Status column to find out when the content was last crawled and when the next crawl is due.
5. Click Cancel to stop a crawl job that is Running or Scheduled.

Rerun a Search Crawl Job

You can rerun a crawl with the status of Terminated or that displays progress totals of zero.

1. In the Oracle Analytics home page, click Navigator , then click Console.
2. In the Console under Configuration and Settings, click Search Index.
3. In the Data Model tab, deselect and then reselect the Enable Data Model Crawl checkbox.
4. Click Save.
5. In the Data Model tab, deselect and then reselect the Enable Data Model Crawl checkbox.
6. Click the Monitor Crawls link and locate the scheduled job. The revised crawl runs in a few minutes time.

Certify a Dataset to Enable Users to Search It from the Home Page

You certify a dataset uploaded by a user so that other users can search it from the home page using the search bar.

As an administrator, you use certification to control how much compute time is consumed by indexing datasets, which can affect system performance. If your dataset is not already indexed, see Index a Dataset.

1. In the Oracle Analytics home page, click Navigator , then click Console.
2. Click Datasets.
3. Hover over the dataset you’d like to certify, click Options , then click Inspect.
   If you can't see Options, expand the size of your browser or scroll to the right-hand side of your device screen.
4. On the General tab, click Certify.
5. Click Save.