Tokenization in OpenSearch

Keep in mind that OpenSearch, by default, tokenizes or analyzes fields that are of type text. As a result, strings consisting of two or more words may become separate fields when connected to Composer (for example, city names like Las Vegas). To disable this process and ensure that a string field is not analyzed, specify its type as keyword:

To learn more about tokenization in Elasticsearch, see Get Trained Models API.

Important: If you are connecting to OpenSearch versions earlier than 2.x, use the Elasticsearch 7 connector.