Azure Vector Search

Question

Dear Community,

I am new to the Azure business which is why I kindly ask for your patience.

Recently, I created I chatbot including the RAG approach. Everything worked perfectly, i.e. I could upload my data on the azure AI platform, selected the necessary resources and embedding tools, told the system to chunk and vectorize it and voila, I could search my own data without a problem. The bot would provide me with the necessary answers as well as the necessary references.

Yesterday this approach stopped working. In spite doing all the same steps as above, I got the following error message:

"An error occurred when calling Azure Cognitive Search: Please assign a proper column/field for vector search. It should be of type Collection(Edm.Single)".

And so I did. In the index for my data, I added such a column, including the necessary vectorizor and so forth.

grafik

However, this did not change anything. While I could ask the bot to search my data, it was only on the basis of keywords, as no vector was provided when I ask the data manually. More importantly, the information tells me that no vector data is provided, see below:

grafik

Your help would be much appreciated.
Thanks and all the best to you.

Fabian

Answer

@Fabian Tomaschek Thanks for asking question!

You may want to know that the "fieldMapping" property is necessary for vector embeddings.

I suspect this requirement arises because you are using two vector fields. According to the documentation, this property is typically only required for Azure Cosmos DB/MongoDB.

Below is the property format you may try to get the request working:

"fieldsMapping": {
                  "vectorFields": [
                      "titleVector",
                      "contentVector"
                  ]
              },

Refer- https://video2.skills-academy.com/en-us/azure/search/vector-search-how-to-create-index?tabs=config-2023-11-01%2Crest-2023-11-01%2Cpush%2Cportal-check-index#add-a-vector-field-to-the-fields-collection

Please let me know.

Share via

Azure Vector Search

1 answer