Relevance Generative Answering (RGA) implementation overview

This is for:

Developer System Administrator

Step 1: Choose the content to use

The content that you choose for RGA will be used as the raw data from which answers are generated. Therefore, the quality of the content has a direct impact on the quality of the answers.

See Relevance Generative Answering (RGA) content requirements and best practices.

Note

An optimal RGA implementation includes both an RGA model and a Semantic Encoder (SE) model. For best results, both models should be configured to use the same content.

See RGA overview for information on how RGA and SE work together in the context of a search session to generate answers.

Step 2: Create an RGA model

An RGA model creates embeddings for the content that you specify in the model settings. The RGA model then uses the embedding vector space to find the most relevant segments of text (chunks) that will be used to generate the answer.

See Create and manage Relevance Generative Answering (RGA) models.

Step 3: Associate the RGA model with a query pipeline

When an RGA model is associated with your search interface’s query pipeline, the model is used to generate answers for queries that are submitted in the search interface.

See Associate a Relevance Generative Answering (RGA) model with a query pipeline.

Important

The RGA model must be associated to the same query pipeline as the Semantic Encoder (SE) model.

Step 4: Create a Semantic Encoder (SE) model

An SE model adds vector-based search capabilities to a Coveo-powered search interface that uses RGA. Vector search allows your search interface to find items based on semantic similarity and not just keyword matches. When a query is submitted in a Coveo-powered search interface, the top search results are sent to the RGA model. The RGA model references only these top search result items to find the most relevant segments of text from which to generate the answer. An SE model ensures that RGA always uses the most relevant content to generate answers.

See Create and manage Semantic Encoder (SE) models.

Note

An optimal RGA implementation includes both an RGA model and a Semantic Encoder (SE) model. For best results, both models should be configured to use the same content.

See RGA overview for information on how RGA and SE work together in the context of a search session to generate answers.

Step 5: Associate the SE model with a query pipeline

When an SE model is associated with your search interface’s query pipeline, the SE model is used to add vector-based search capabilities.

See Associate a Semantic Encoder (SE) model with a query pipeline.

Important

The SE model must be associated to the same query pipeline as the RGA model.

Step 6: Configure a search interface for RGA

To add the RGA question-answering experience to your Coveo-powered search interface, you must enable the RGA component.

See Configure a search interface for Relevance Generative Answering (RGA).

Step 7: Add a custom dimension to report on RGA events

To track the usage of Relevance Generative Answering (RGA), Coveo Usage Analytics (Coveo UA) logs events for RGA and the actions performed on the RGA component in your search interface. To report on these events, you must add a custom dimension to your Coveo organization.

See Relevance Generative Answering (RGA) reports and UA events.