semantic reranker of azureaisearch

[aradhanachaturvedi]

I want to use semantic reranker of azure ai search to search from Azure AI search vector store and use threshold to use most relevant documents. How can I use search.reranker score?

[dosubot[bot]]

To use the search.reranker score with Azure AI Search and apply a threshold for the most relevant documents, you can utilize the semantic_hybrid_search_with_score method with the score_type set to "reranker_score". You can also specify a score_threshold to filter out documents that do not meet the relevance criteria. Here's an example of how you can implement this:

def search_with_reranker_score(query: str, k: int, score_threshold: float):
    # Assuming `azure_search` is an instance of AzureSearch
    results = azure_search.semantic_hybrid_search_with_score(
        query=query,
        k=k,
        score_type="reranker_score",
        score_threshold=score_threshold
    )
    return results

This function will return a list of documents and their reranker scores, filtered by the specified threshold ^[1].

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Bug Report | Other}

[aradhanachaturvedi]

@dosu. But I want to use search reranker score .which score is it???How do I print and view

[dosubot[bot]]

To print and view the search reranker score in Azure AI Search, you can extract it from the search results using the key @search.reranker_score. Here's a code snippet that demonstrates how to access and print the reranker score for each document in the search results:

docs = [
    (
        Document(
            page_content=result.pop(FIELDS_CONTENT),
            metadata={
                **(
                    {FIELDS_ID: result.pop(FIELDS_ID)}
                    if FIELDS_ID in result
                    else {}
                ),
                **(
                    json.loads(result[FIELDS_METADATA])
                    if FIELDS_METADATA in result
                    else {
                        k: v
                        for k, v in result.items()
                        if k != FIELDS_CONTENT_VECTOR
                    }
                ),
                **{
                    "captions": (
                        {
                            "text": result.get("@search.captions", [{}])[0].text,
                            "highlights": result.get("@search.captions", [{}])[0].highlights,
                        }
                        if result.get("@search.captions")
                        else {}
                    ),
                    "answers": semantic_answers_dict.get(
                        result.get(FIELDS_ID, ""),
                        "",
                    ),
                },
            },
        ),
        float(result["@search.score"]),
        float(result["@search.reranker_score"]),
    )
    for result in results
]

# Print reranker scores
for doc, score, reranker_score in docs:
    print(f"Document: {doc}, Reranker Score: {reranker_score}")

In this code, float(result["@search.reranker_score"]) is used to extract and convert the reranker score to a float for each document. You can then print this score as needed ^[1].

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Bug Report | Other}