Using my own score function to sort results

Elastic Search
1

I have a list of documents, which contain basic sentences. Depending on the semantic similarity to a search query, I want to give each sentence a score. For this purpose, I am using a neural network, which will do exactly that. It takes a query sentence as input, transforms it to a vector and computes the cosine similarity with all the sentences in my database, which represents a similarity score. I want to then sort my results based on that similarity score.

I am therefore looking for a way, to implement my personal scoring function into the App Search Environment. How can I do that?

Thank you for your help!

2

Hello — App Search does allow customization of scoring via the dashboard and the API, but not at the level you are looking for.

Elasticsearch can accomplish what you describe. Here's a great post about it: https://www.elastic.co/blog/text-similarity-search-with-vectors-in-elasticsearch

3

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.