How to perform keyword search with Azure AI search

Question

Hi, I am able to successfully run vector and hybrid search on Azure AI Search following this example:
https://github.com/Azure/azure-sdk-for-python/blob/main/sdk/search/azure-search-documents/samples/sample_vector_search.py I would like to perform pure keyword search. For that I have added the following function to the code above:

def keyword_search():
    # [START single_vector_search]
    query = "n/a"

    search_client = SearchClient(service_endpoint, index_name, AzureKeyCredential(key))

    results = search_client.search(
        search_text=query,
        vector_queries=None,
        select=["hotelId", "hotelName"],
    )

    for result in results:
        print(result)
    # [END single_vector_search]

The code runs and returns the following result:
{'hotelId': '2', 'hotelName': 'Roach Motel', '@search.score': 1.242726, '@search.reranker_score': None, '@search.highlights': None, '@search.captions': None}

But the query 'n/a' doesn't appear in the data and shouldn't match any result. The results variable should be empty.

In other words, how can I ensure that search() does match at least one keyword when it search_text is not None and vector_queries is None.

Thanks for your help.

Answer 1

It seems that "n/a" is tokenized internally into "n a" and therefore matching these letters