What is the best way to rank external API data and internal data from OpenSearch?

There could be multiple ways using which you will get desired output. However, for the most relevant ranking you have to test out patterns. However, for Relevance-based score calculation You can use sort clauses with relevant scoring . Also, as OpenSearch assign relevance scores using keyword, document age, algorithm or any custom rule. You need to assign relevance score appropriately for both your internal and external data, you can then sort the combined results based on their relevance scores.

You can refer to the below documentation for detailed information on "Learning to Rank" best-practices for Amazon OpenSearch Service.

https://docs.aws.amazon.com/opensearch-service/latest/developerguide/learning-to-rank.html