Support semantic reranking using contextual snippets instead of entire field text

[kderusso]

Followup to the POC described in #128255

Adds the ability to rerank based on a smaller number of snippets.

Example, with a default of one snippet:

GET test/_search
{
  "retriever": {
    "text_similarity_reranker": {
      "retriever": {
        "standard": {
          "query": {
            "term": {
              "other": "lotr"
            }
          }
        }
      },
      "rank_window_size": 2,
      "field": "semantic_text",
      "inference_text": "are all who wander lost?",
      "snippets": { }
    }
  }
}

Example, specifying snippets:

GET test/_search
{
  "retriever": {
    "text_similarity_reranker": {
      "retriever": {
        "standard": {
          "query": {
            "term": {
              "other": "lotr"
            }
          }
        }
      },
      "rank_window_size": 2,
      "field": "semantic_text",
      "inference_text": "are all who wander lost?",
      "snippets": {
        "num_snippets": 3
      }
    }
  }
}

Not specifying snippets will continue to send the entire field contents into the reranker model.

[cla-checker-service]

💚 CLA has been signed

[kderusso]

cla/check

[jimczi]

Do we really need the separation here between TextSimilarityRerankingRankFeaturePhaseRankShardContext and RerankingRankFeaturePhaseRankShardContext? I don't see RerankingRankFeaturePhaseRankShardContext being used elsewhere so maybe we can just have TextSimilarityRerankingRankFeaturePhaseRankShardContext and implement the full logic there?

[kderusso]

I agree that it's a little extra, but it clearly conveys the fact that snippets only pertain to text similarity reranking and not other rerankers we may add in the future. If you feel strongly about it I will refactor back, but I don't think this is a bad thing per se.

[jimczi]

Can we just rename RerankingRankFeaturePhaseRankShardContext into TextSimilarityRerankingRankFeaturePhaseRankShardContext? I don't see what RerankingRankFeaturePhaseRankShardContext provides as an abstract class.

[kderusso]

Can we just rename RerankingRankFeaturePhaseRankShardContext into TextSimilarityRerankingRankFeaturePhaseRankShardContext? I don't see what RerankingRankFeaturePhaseRankShardContext provides as an abstract class.

@jimczi this class is referenced elsewhere including the random reranker. I personally think it's cleaner to have the separation, but if you feel strongly about it I will put snippet configuration in the parent class and remove the abstraction. I am commenting one more time instead of changing it, because I actually prefer this separation to putting things specific to text similarity that are used by the parent reranker. However, I won't die on that hill if you absolutely hate it.

[jimczi]

Is the separation between SnippetRankInput and RerankSnippetConfig necessary? Why not injecting numSnippets and snippetQueryBuilder directly here?

[kderusso]

It's a little convoluted because we don't allow size of snippets in the API, I'll see if I can clean it up

[kderusso]

Updated in d56726c

[kderusso]

@jimczi Here is the TODO I referenced in my earlier response 🙂

[jimczi]

Thanks for adding the feature flag. I left some minor comments to address but the feature looks good to me.

[jimczi]

Can we just rename RerankingRankFeaturePhaseRankShardContext into TextSimilarityRerankingRankFeaturePhaseRankShardContext? I don't see what RerankingRankFeaturePhaseRankShardContext provides as an abstract class.

Notr relevant