Perform semantic search and retrieval-augmented generation

This tutorial guides you through the end-to-end process of creating and using text embeddings for semantic search and retrieval-augmented generation (RAG).

This tutorial covers the following tasks:

• Creating a BigQuery ML remote model over a Vertex AI embedding model.
• Using the remote model with the ML.GENERATE_EMBEDDING function to generate embeddings from text in a BigQuery table.
• Creating a vector index to index the embeddings in order to improve search performance.
• Using the VECTOR_SEARCH function with the embeddings to search for similar text.
• Perform RAG by generating text with the ML.GENERATE_TEXT function, and using vector search results to augment the prompt input and improve results.

This tutorial uses the BigQuery public table patents-public-data.google_patents_research.publications.

Required roles and permissions

• To create a connection, you need membership in the following Identity and Access Management (IAM) role:

  • roles/bigquery.connectionAdmin

• To grant permissions to the connection's service account, you need the following permission:

  • resourcemanager.projects.setIamPolicy

• The IAM permissions needed in this tutorial for the remaining BigQuery operations are included in the following two roles:

  • BigQuery Data Editor (roles/bigquery.dataEditor) to create models, tables, and indexes.
  • BigQuery User (roles/bigquery.user) to run BigQuery jobs.

Costs

In this document, you use the following billable components of Google Cloud:

• BigQuery ML: You incur costs for the data that you process in BigQuery.
• Vertex AI: You incur costs for calls to the Vertex AI service that's represented by the remote model.

To generate a cost estimate based on your projected usage, use the pricing calculator. New Google Cloud users might be eligible for a free trial.

For more information about BigQuery pricing, see BigQuery pricing in the BigQuery documentation.

For more information about Vertex AI pricing, see the Vertex AI pricing page.

Before you begin

1. In the Google Cloud console, on the project selector page, select or create a Google Cloud project.

   Note: If you don't plan to keep the resources that you create in this procedure, create a project instead of selecting an existing project. After you finish these steps, you can delete the project, removing all resources associated with the project.

   Go to project selector

2. Make sure that billing is enabled for your Google Cloud project.

3. Enable the BigQuery, BigQuery Connection, and Vertex AI APIs.

   Enable the APIs

Create a dataset

Create a BigQuery dataset to store your ML model.

Console

1. In the Google Cloud console, go to the BigQuery page.

   Go to the BigQuery page

2. In the Explorer pane, click your project name.

3. Click View actions > Create dataset.

   

4. On the Create dataset page, do the following:

   • For Dataset ID, enter bqml_tutorial.

   • For Location type, select Multi-region, and then select US (multiple regions in United States).

   • Leave the remaining default settings as they are, and click Create dataset.

bq

To create a new dataset, use the bq mk command with the --location flag. For a full list of possible parameters, see the bq mk --dataset command reference.

1. Create a dataset named bqml_tutorial with the data location set to US and a description of BigQuery ML tutorial dataset:

   bq --location=US mk -d \
    --description "BigQuery ML tutorial dataset." \
    bqml_tutorial

   Instead of using the --dataset flag, the command uses the -d shortcut. If you omit -d and --dataset, the command defaults to creating a dataset.

2. Confirm that the dataset was created:

   bq ls

API

Call the datasets.insert method with a defined dataset resource.

{
  "datasetReference": {
     "datasetId": "bqml_tutorial"
  }
}

BigQuery DataFrames

Before trying this sample, follow the BigQuery DataFrames setup instructions in the BigQuery quickstart using BigQuery DataFrames. For more information, see the BigQuery DataFrames reference documentation.

To authenticate to BigQuery, set up Application Default Credentials. For more information, see Set up ADC for a local development environment.

import google.cloud.bigquery

bqclient = google.cloud.bigquery.Client()
bqclient.create_dataset("bqml_tutorial", exists_ok=True)

Create a connection

Create a Cloud resource connection and get the connection's service account. Create the connection in the same location as the dataset you created in the previous step.

You can skip this step if you either have a default connection configured, or you have the BigQuery Admin role.

Create a Cloud resource connection for the remote model to use, and get the connection's service account. Create the connection in the same location as the dataset that you created in the previous step.

Select one of the following options:

Console

1. Go to the BigQuery page.

   Go to BigQuery

2. In the Explorer pane, click Add data:

   

   The Add data dialog opens.

3. In the Filter By pane, in the Data Source Type section, select Business Applications.

   Alternatively, in the Search for data sources field, you can enter Vertex AI.

4. In the Featured data sources section, click Vertex AI.

5. Click the Vertex AI Models: BigQuery Federation solution card.

6. In the Connection type list, select Vertex AI remote models, remote functions and BigLake (Cloud Resource).

7. In the Connection ID field, enter a name for your connection.

8. Click Create connection.

9. Click Go to connection.

10. In the Connection info pane, copy the service account ID for use in a later step.

bq

1. In a command-line environment, create a connection:

   bq mk --connection --location=REGION --project_id=PROJECT_ID \
       --connection_type=CLOUD_RESOURCE CONNECTION_ID

   The --project_id parameter overrides the default project.

   Replace the following:

   • REGION: your connection region
   • PROJECT_ID: your Google Cloud project ID
   • CONNECTION_ID: an ID for your connection

   When you create a connection resource, BigQuery creates a unique system service account and associates it with the connection.

   Troubleshooting: If you get the following connection error, update the Google Cloud SDK:

   Flags parsing error: flag --connection_type=CLOUD_RESOURCE: value should be one of...
   

2. Retrieve and copy the service account ID for use in a later step:

   bq show --connection PROJECT_ID.REGION.CONNECTION_ID

   The output is similar to the following:

   name                          properties
   1234.REGION.CONNECTION_ID     {"serviceAccountId": "[e-mail]"}
   

Terraform

Use the google_bigquery_connection resource.

Note: To create BigQuery objects using Terraform, you must enable the Cloud Resource Manager API.

To authenticate to BigQuery, set up Application Default Credentials. For more information, see Set up authentication for client libraries.

The following example creates a Cloud resource connection named my_cloud_resource_connection in the US region:

# This queries the provider for project information.
data "google_project" "default" {}

# This creates a cloud resource connection in the US region named my_cloud_resource_connection.
# Note: The cloud resource nested object has only one output field - serviceAccountId.
resource "google_bigquery_connection" "default" {
  connection_id = "my_cloud_resource_connection"
  project       = data.google_project.default.project_id
  location      = "US"
  cloud_resource {}
}

To apply your Terraform configuration in a Google Cloud project, complete the steps in the following sections.

Prepare Cloud Shell

1. Launch Cloud Shell.

2. Set the default Google Cloud project where you want to apply your Terraform configurations.

   You only need to run this command once per project, and you can run it in any directory.

   export GOOGLE_CLOUD_PROJECT=PROJECT_ID

   Environment variables are overridden if you set explicit values in the Terraform configuration file.

Prepare the directory

Each Terraform configuration file must have its own directory (also called a root module).

1. In Cloud Shell, create a directory and a new file within that directory. The filename must have the .tf extension—for example main.tf. In this tutorial, the file is referred to as main.tf.

   mkdir DIRECTORY && cd DIRECTORY && touch main.tf

2. If you are following a tutorial, you can copy the sample code in each section or step.

   Copy the sample code into the newly created main.tf.

   Optionally, copy the code from GitHub. This is recommended when the Terraform snippet is part of an end-to-end solution.

3. Review and modify the sample parameters to apply to your environment.
4. Save your changes.
5. Initialize Terraform. You only need to do this once per directory.

   terraform init

   Optionally, to use the latest Google provider version, include the -upgrade option:

   terraform init -upgrade

Apply the changes

1. Review the configuration and verify that the resources that Terraform is going to create or update match your expectations:

   terraform plan

   Make corrections to the configuration as necessary.

2. Apply the Terraform configuration by running the following command and entering yes at the prompt:

   terraform apply

   Wait until Terraform displays the "Apply complete!" message.

3. Open your Google Cloud project to view the results. In the Google Cloud console, navigate to your resources in the UI to make sure that Terraform has created or updated them.

Note: Terraform samples typically assume that the required APIs are enabled in your Google Cloud project.

Grant the service account access

Grant the connection's service account the Vertex AI User role. You must grant this role in the same project you created or selected in the Before you begin section. Granting the role in a different project results in the error [e-mail] does not have the permission to access resource.

To grant the role, follow these steps:

1. Go to the IAM & Admin page.

   Go to IAM & Admin

2. Click Grant Access.

3. In the New principals field, enter the service account ID that you copied earlier.

4. In the Select a role field, choose Vertex AI, and then select Vertex AI User role.

5. Click Save.

Create the remote model for text embedding generation

Create a remote model that represents a hosted Vertex AI text embedding generation model:

1. In the Google Cloud console, go to the BigQuery page.

   Go to BigQuery

2. In the query editor, run the following statement:

   CREATE OR REPLACE MODEL `bqml_tutorial.embedding_model`
     REMOTE WITH CONNECTION `us.CONNECTION_ID`
     OPTIONS (ENDPOINT = 'text-embedding-005');

   Replace the CONNECTION_ID with the ID of your BigQuery connection.

   When you view the connection details in the Google Cloud console, the CONNECTION_ID is the value in the last section of the fully qualified connection ID that is shown in Connection ID, for example projects/myproject/locations/connection_location/connections/myconnection.

   The query takes several seconds to complete, after which the model embedding_model appears in the bqml_tutorial dataset in the Explorer pane. Because the query uses a CREATE MODEL statement to create a model, there are no query results.

Generate text embeddings

Generate text embeddings from patent abstracts using the ML.GENERATE_EMBEDDING function, and then write them to a BigQuery table so that they can be searched.

1. In the Google Cloud console, go to the BigQuery page.

   Go to BigQuery

2. In the query editor, run the following statement:

   CREATE OR REPLACE TABLE `bqml_tutorial.embeddings` AS
   SELECT * FROM ML.GENERATE_EMBEDDING(
     MODEL `bqml_tutorial.embedding_model`,
     (
       SELECT *, abstract AS content
       FROM `patents-public-data.google_patents_research.publications`
       WHERE LENGTH(abstract) > 0 AND LENGTH(title) > 0 AND country = 'Singapore'
     )
   )
   WHERE LENGTH(ml_generate_embedding_status) = 0;

This query takes approximately 5 minutes to complete.

Embedding generation using the ML.GENERATE_EMBEDDING function might fail due to Vertex AI LLM quotas or service unavailability. Error details are returned in the ml_generate_embedding_status column. An empty ml_generate_embedding_status column indicates successful embedding generation.

For alternative text embedding generation methods in BigQuery, see the Embed text with pretrained TensorFlow models tutorial.

Create a vector index

If you create a vector index on an embedding column, a vector search performed on that column uses the Approximate Nearest Neighbor search technique. This technique improves vector search performance, with the trade-off of reducing recall and so returning more approximate results.

To create a vector index, use the CREATE VECTOR INDEX data definition language (DDL) statement:

1. Go to the BigQuery page.

   Go to BigQuery

2. In the query editor, run the following SQL statement:

   CREATE OR REPLACE VECTOR INDEX my_index
   ON `bqml_tutorial.embeddings`(ml_generate_embedding_result)
   OPTIONS(index_type = 'IVF',
     distance_type = 'COSINE',
     ivf_options = '{"num_lists":500}')

Creating a vector index typically takes only a few seconds. It takes another 2 or 3 minutes for the vector index to be populated and ready to use.

Verify vector index readiness

The vector index is populated asynchronously. You can check whether the index is ready to be used by querying the INFORMATION_SCHEMA.VECTOR_INDEXES view and verifying that the coverage_percentage column value is greater than 0 and the last_refresh_time column value isn't NULL.

1. Go to the BigQuery page.

   Go to BigQuery

2. In the query editor, run the following SQL statement:

   SELECT table_name, index_name, index_status,
   coverage_percentage, last_refresh_time, disable_reason
   FROM `PROJECT_ID.bqml_tutorial.INFORMATION_SCHEMA.VECTOR_INDEXES`

   Replace PROJECT_ID with your project ID.

Perform a text similarity search using the vector index

Use the VECTOR_SEARCH function to search for relevant patents that match embeddings generated from a text query.

The top_k argument determines the number of matches to return, in this case five. The fraction_lists_to_search option determines the percentage of vector index lists to search. The vector index you created has 500 lists, so the fraction_lists_to_search value of .01 indicates that this vector search scans five of those lists. A lower fraction_lists_to_search value as shown here provides lower recall and faster performance. For more information about vector index lists, see the num_lists vector index option.

The model you use to generate the embeddings in this query must be the same as the one you use to generate the embeddings in the table you are comparing against, otherwise the search results won't be accurate.

1. Go to the BigQuery page.

   Go to BigQuery

2. In the query editor, run the following SQL statement:

   SELECT query.query, base.publication_number, base.title, base.abstract
   FROM VECTOR_SEARCH(
     TABLE `bqml_tutorial.embeddings`, 'ml_generate_embedding_result',
     (
     SELECT ml_generate_embedding_result, content AS query
     FROM ML.GENERATE_EMBEDDING(
     MODEL `bqml_tutorial.embedding_model`,
     (SELECT 'improving password security' AS content))
     ),
     top_k => 5, options => '{"fraction_lists_to_search": 0.01}')

   The output is similar to the following:

   +-----------------------------+--------------------+-------------------------------------------------+-------------------------------------------------+
   |            query            | publication_number |                       title                     |                      abstract                   |
   +-----------------------------+--------------------+-------------------------------------------------+-------------------------------------------------+
   | improving password security | SG-120868-A1       | Data storage device security method and a...    | Methods for improving security in data stora... |
   | improving password security | SG-10201610585W-A  | Passsword management system and process...      | PASSSWORD MANAGEMENT SYSTEM AND PROCESS ...     |
   | improving password security | SG-148888-A1       | Improved system and method for...               | IMPROVED SYSTEM AND METHOD FOR RANDOM...        |
   | improving password security | SG-194267-A1       | Method and system for protecting a password...  | A system for providing security for a...        |
   | improving password security | SG-120868-A1       | Data storage device security...                 | Methods for improving security in data...       |
   +-----------------------------+--------------------+-------------------------------------------------+-------------------------------------------------+
   

Create the remote model for text generation

Create a remote model that represents a hosted Vertex AI text generation model:

1. In the Google Cloud console, go to the BigQuery page.

   Go to BigQuery

2. In the query editor, run the following statement:

   CREATE OR REPLACE MODEL `bqml_tutorial.text_model`
     REMOTE WITH CONNECTION `us.CONNECTION_ID`
     OPTIONS (ENDPOINT = 'gemini-1.5-flash-002');

   Replace the CONNECTION_ID with the ID of your BigQuery connection.

   When you view the connection details in the Google Cloud console, the CONNECTION_ID is the value in the last section of the fully qualified connection ID that is shown in Connection ID, for example projects/myproject/locations/connection_location/connections/myconnection.

   The query takes several seconds to complete, after which the model text_model appears in the bqml_tutorial dataset in the Explorer pane. Because the query uses a CREATE MODEL statement to create a model, there are no query results.

Generate text augmented by vector search results

Feed the search results as prompts to generate text with the ML.GENERATE_TEXT function

1. In the Google Cloud console, go to the BigQuery page.

   Go to BigQuery

2. In the query editor, run the following statement:

   SELECT ml_generate_text_llm_result AS generated, prompt
   FROM ML.GENERATE_TEXT(
     MODEL `bqml_tutorial.text_model`,
     (
       SELECT CONCAT(
         'Propose some project ideas to improve user password security using the context below: ',
         STRING_AGG(
           FORMAT("patent title: %s, patent abstract: %s", base.title, base.abstract),
           ',\n')
         ) AS prompt,
       FROM VECTOR_SEARCH(
         TABLE `bqml_tutorial.embeddings`, 'ml_generate_embedding_result',
         (
           SELECT ml_generate_embedding_result, content AS query
           FROM ML.GENERATE_EMBEDDING(
             MODEL `bqml_tutorial.embedding_model`,
            (SELECT 'improving password security' AS content)
           )
         ),
       top_k => 5, options => '{"fraction_lists_to_search": 0.01}')
     ),
     STRUCT(600 AS max_output_tokens, TRUE AS flatten_json_output));

   The output is similar to the following:

   +------------------------------------------------+------------------------------------------------------------+
   |            generated                           | prompt                                                     |
   +------------------------------------------------+------------------------------------------------------------+
   | These patents suggest several project ideas to | Propose some project ideas to improve user password        |
   | improve user password security.  Here are      | security using the context below: patent title: Active     |
   | some, categorized by the patent they build     | new password entry dialog with compact visual indication   |
   | upon:                                          | of adherence to password policy, patent abstract:          |
   |                                                | An active new password entry dialog provides a compact     |
   | **I. Projects based on "Active new password    | visual indication of adherence to password policies. A     |
   | entry dialog with compact visual indication of | visual indication of progress towards meeting all          |
   | adherence to password policy":**               | applicable password policies is included in the display    |
   |                                                | and updated as new password characters are being...        |
   +------------------------------------------------+------------------------------------------------------------+
    

Clean up

Caution: Deleting a project has the following effects:

• Everything in the project is deleted. If you used an existing project for the tasks in this document, when you delete it, you also delete any other work you've done in the project.
• Custom project IDs are lost. When you created this project, you might have created a custom project ID that you want to use in the future. To preserve the URLs that use the project ID, such as an appspot.com URL, delete selected resources inside the project instead of deleting the whole project.

If you plan to explore multiple architectures, tutorials, or quickstarts, reusing projects can help you avoid exceeding project quota limits.

1. In the Google Cloud console, go to the Manage resources page.

   Go to Manage resources

2. In the project list, select the project that you want to delete, and then click Delete.
3. In the dialog, type the project ID, and then click Shut down to delete the project.

What's next

• Try the Parse PDFs in a retrieval-augmented generation pipeline tutorial to learn how to create a RAG pipeline based on parsed PDF content.