Responses for Azure Openai Web app

Question

Hi,

I have built an Azure Openai web app for my custom data. I am facing issues with the responses. Sometimes the openai response is fast, sometime it doesn't work, there is no consistency in responses. Also its not picking all the data that a query should match with. How can I deal with it. It doesn't show all the citations, sometimes it doesn't show any citation.

I have one more question, if 100 documents are uploaded, can it provide me the names or title of all the documents from same index?

It would be helpful if you suggest how to deal with structure data in cognitive search with the unstructure data in the same index.

Answer 1

@Deepak Dange

Thanks for reaching here! It seems like you are facing some issues with the consistency of responses from Azure OpenAI service. There could be several reasons for this, such as network latency, server load, or the complexity of your queries.

To improve the consistency of responses, you can try the following:

• Optimize your queries: Make sure that your queries are well-structured and optimized for performance. You can use query profiling to identify slow queries and optimize them.
• Use caching: You can use caching to store frequently accessed data in memory, which can improve the response time of your application.
• Monitor server load: Keep an eye on the server load and scale your resources accordingly. You can use Azure Monitor to monitor the performance of your application and scale your resources based on the load.

Regarding your second question, you can retrieve the names or titles of all the documents from the same index. You can use the $select parameter to specify the fields that you want to retrieve. For example, to retrieve the names or titles of all the documents, you can use the following query:

https://[service name].search.windows.net/indexes/[index name]/docs?api-version=[api-version]&$select=name

This will return a list of all the names or titles of the documents in the index.

Finally, to deal with structured and unstructured data in the same index, you can use field mappings to map the fields in your unstructured data to the fields in your structured data.

This will allow you to search and retrieve data from both types of fields using a single query. You can also use cognitive skills to extract structured data from unstructured data during indexing.

You may refer to below links- see- https://learn.microsoft.com/en-us/azure/search/tutorial-multiple-data-sources

https://learn.microsoft.com/en-us/azure/search/search-indexer-field-mappings?tabs=rest

https://learn.microsoft.com/en-us/azure/search/cognitive-search-skill-document-extraction

Hope this helps let us know.