Indexing in elasticsearch

[homework36]

Hi,

I found that Elasticsearch has disabled the JSON-LD plugin. This means that we can no longer send compact JSON-LD to the server and have it expanded on the server. Instead, we can only expand JSON-LD first and then send it to the server. However, there's an important issue related to this approach. Here's a simplified example from CantusDB:

[
    {
      "@id": "https://cantusdatabase.org/chant/562072",
      "@type": [
        "http://www.wikidata.org/entity/Q23072435"
      ],
      "http://www.wikidata.org/entity/Q4484726": [
        {
          "@value": "G"
        }
      ]
    }
]

In this example, the property name "finalis" is expanded to a link. During the indexing process, Elasticsearch sanitizes this link into text (and it has to be done this way. We might be able to do type keyword but it won't be much different). This means if we want to run a query like "finalis=F," we must use "[the long link]=F," which is almost useless. I don't know how to resolve this issue. I did some research, and it seems there are two potential solutions:

1. Store this link separately with "@ id" or something else. (But "@ id" is typically used for a specific object?)
2. Modify the existing JSON-LD data structure and add "@ name" so the mapping will look like this

{
  "mappings": {
    "properties": {
      "@id": {
        "type": "keyword"
      },
      "@type": {
        "type": "keyword"
      },
      "http://www.wikidata.org/entity/Q4484726": {
        "type": "nested",
        "properties": {
          "@value": {
            "type": "keyword"  // Store values as keyword
          },
          "@name": {
            "type": "text"     // Store names as text
          }
        }
      }
    }
  }
}

But this would only enable "finalis=F," and we have to manually add all the names/synonyms, which doesn't seem to fit our goal...

I might misunderstand some features regarding jsonld and linked data, so any help or idea is very much appreciated!