I have been using azure ai search and scoring profiles to boost the documents of my index that come form the ‘reviewed’ source that means I want to send to the very TOP documents that have the string ‘reviewed’ on the field source, so I configured this scoring profile:

  "scoringProfiles": [
    {
      "name": "pubs_peer_house",
      "functionAggregation": "sum",
      "text": {
        "weights": {
          "title": 3,
          "content": 10
        }
      },
      "functions": [
        {
          "fieldName": "source",
          "interpolation": "linear",
          "type": "tag",
          "boost": 100,
          "freshness": null,
          "magnitude": null,
          "distance": null,
          "tag": {
            "tagsParameter": "reviewed"
          }
        }
      ]
    }
  ],

I have this code in python:

from azure.core.exceptions import HttpResponseError

# Define the search query
search_query = "cybersecurity in Arizona from tech today"

# Function to get the embedding of the search query
search_vector = get_embedding(search_query)

# Define the scoring parameters
scoring_parameters = ["reviewed-100"]  

try:
    # Perform the search using the Azure Cognitive Search client
    response = search_client.search(
        search_query, 
        top=5,  # Return the top 5 results
        vector_queries=[
            VectorizedQuery(
                vector=search_vector,  # The vector representation of the search query
                k_nearest_neighbors=5,  # Number of nearest neighbors to find
                fields="vector_space_search"  # Field in the index to search the vector space
            )
        ],
        query_type="semantic",  # Specify the query type as semantic
        semantic_configuration_name=semantic_profile,  # The name of the semantic configuration
        scoring_profile='pubs_peer_house',  # The scoring profile to use
        scoring_parameters=scoring_parameters  # The scoring parameters
    )

    # Iterate over the search results
    for idx, doc in enumerate(response, start=1):
        # Default values if fields are not found
        found_content = "Not found"
        
        # Extract fields from the search result
        date = doc.get('date', 'N/A')  # Date of the document
        source = doc.get('source', 'N/A')  # Source of the document
        title = doc.get('title', 'N/A')  # Title of the document
        content = doc.get('content', 'N/A')  # Content of the document
        
        # Print the results
        print(f"{idx}")
        print(f"Score: {doc['@search.score']:.5f}")
        print(f"Source: {source}")
        print(f"Title: {title}")
        print(f"Content: {content}nn")
except HttpResponseError as e:
    # Handle HTTP response errors
    print(f"HTTP Response Error: {e.message}")
    print(f"Details: {e.response}")
except Exception as ex:
    # Handle other exceptions
    print(f"An error occurred: {ex}")

Nontheless when I ask for anything and I implemente my profiles scoring along with the semantic ranker in a hybrid search it doesnt matter the value of the booster I always get the same results

look:

1 Score: 0.03667 Source: americas Subtitle: What level of
cybersecurity do you have? Content: We comply with industry standards
for cybersecurity and recommend that you…

2 Score: 0.02639 Source: reviewed Subtitle: What do I need to operate
securely? Content: The key or security signature. It is an 8-digit
alphanumeric code ML te….

3 Score: 0.01562 Source: europe Subtitle: Passkey password and pin
still better than faceID Content: careful whose face do you trust….

even with params like : scoring_parameters = ["reviewed-2500000"] I still get:

1 Score: 0.03667
Source: americas
Subtitle: What level of
cybersecurity do you have? Content: We comply with industry standards
for cybersecurity and recommend that you…

2 Score: 0.02639 Source: reviewed Subtitle: What do I need to operate
securely? Content: The key or security signature. It is an 8-digit
alphanumeric code ML te….

3 Score: 0.01562 Source: europe Subtitle: Passkey password and pin
still better than faceID Content: careful whose face do you trust….

Am I doing something wrong I cant seem to find a tutorial on this in python online. Thank you so much for all of your help guys.