How to get the content recommendation based on the user chat history in a AOAI chat webapp

Question

We have a AOAI chat webapp, using azure open ai service and azure cognitive search, please suggest a way to provide content recomendation to the user based on his/her chat history in the current session.

Answer 1

Hi @TagnaviMY-1268 to provide content recommendations based on the user’s chat history in an Azure OpenAI (AOAI) chat webapp, you can leverage a combination of Azure Cognitive Search, Azure Cosmos DB, and Azure Machine Learning. Here’s a high-level approach:

1. You can enable chat history in your AOAI chat webapp. This feature provides your users with access to their previous queries and responses, allowing them to easily reference past conversations. You can store this chat history in Azure Cosmos DB
2. Analyzing past interactions provides valuable insights into user behavior, preferences, and recurring issues. Armed with this data, you can make informed decisions to optimize user experiences, tailor content, and refine your application’s performance
3. Azure Cognitive Search can be used to index the chat history stored in Azure Cosmos DB. You can create a scoring profile in your index schema and assign weights to the fields you want to prioritize Use the scoring profile in your search queries to show suggested content in response.
4. You can use Azure Machine Learning to train a recommendation model based on the user’s chat history. This model can predict the probability that a user will be interested in an item.
5. Once the model is trained, you can deploy it as an API using Azure Machine Learning3. This API can then be used to provide real-time content recommendations to the user based on their chat history.