Common Functionality or Rules | Teradata Text Analytics AI Functions - Common Functionality or Rules - Teradata Analytics

Use Text Analytics AI functions to conduct text analytics using a model hosted by Azure AI, Google Gemini or AWS Bedrock. Each function will follow a uniform syntax customized for the particular cloud platform.

The query encompasses parameters that ensure a secure authentication process to connect to the cloud platform. Additionally, it specifies the model name along with any relevant model-specific parameters, if applicable.

There are two ways to establish a secure connection to the cloud platform: by using a Teradata authorization object or by providing individual function arguments. Teradata recommends using authorization objects as the preferred method for securely passing credentials.

You can create a Teradata authorization object using the CREATE AUTHORIZATION statement. The object can be defined in the database using the following syntax:

{ CREATE | REPLACE } AUTHORIZATION test_auth
user 'test'
password 'test'
Sesssion_Token 'session_token'

Examples of CREATE AUTHORIZATION Statement

CREATE AUTHORIZATION td_gen_aws_auth
       user '{AWS_ACCESS_KEY}'
       password '{AWS_SECRET_KEY}';
	   Session_Token '{AWS_SESSION_TOKEN}';

CREATE AUTHORIZATION td_gen_azure_auth
       user '{API_BASE}'
       password '{API_KEY}'
       Session_Token '2024-02-15-preview';

CREATE AUTHORIZATION td_gen_gcp_auth
       user '{GCP_PROJECT}'
       password 'us-central1'
       Session_Token '{GCP_ACCESS_TOKEN}';

CREATE AUTHORIZATION td_gen_nim_auth
       user '{API_BASE}'
       password '{API_KEY}';

Mapping of Authorization Object and Text Analytics AI Function Arguments

Cloud Connection Syntax

Use the relevant syntax elements to connect to your cloud platform or instance.

• Azure CSP LLM Syntax
• AWS Bedrock CSP LLM Syntax
• Google Gemini LLM Syntax
• NVIDIA NIM Hosted Models Syntax
• Lite LLM Hosted Models Syntax