Originally posted by **NOT-HAL9000** August 22, 2022
Hey Everyone,
I was having a look through data.gov.au and I found the basic search results to be pretty unusable.
Would it be a bad idea to explore using document embeddings + vector similarity in elastic in order to surface even more relevant results?
See elastic article on it here: [Text Embeddings and Vector Search with elastic](https://www.elastic.co/blog/how-to-deploy-nlp-text-embeddings-and-vector-search)
Discussed in https://github.com/magda-io/magda/discussions/3400