search

opensearch-project / neural-search

Plugin that adds dense neural retrieval into the OpenSearch ecosytem
Apache License 2.0
62 stars 65 forks source link

[PROPOSAL] Add/enhance educational materials related to usage of queries and processors maintained in this repo #948

Open martin-gaievski opened 1 day ago

martin-gaievski commented 1 day ago

What/Why

What are you proposing?

As a user I would like to have access to more documentation and examples of neural search components, like neural/hybrid queries and different processor types. This should help to speedup the ramp-up for new features and provide better self-service in scenarios like "why document X has that low/high score".

What users have asked for this feature?

For example, our main page for hybrid query https://opensearch.org/docs/latest/search-plugins/hybrid-search/#using-hybrid-search does have some examples for weights, but it's not explicitly stated how user can identify those weights in the actual system. Another example would be the details of how scores are compute in hybrid query depending on distribution of documents, number of shards/replicas/data nodes, size of requested documents in sub-query etc. Ideally it can be something visual and interactive, maybe blog (good example is recently published blog https://opensearch.org/blog/How-does-the-rank-normalization-work-in-hybrid-search/).

vibrantvarun commented 1 day ago

Thanks @martin-gaievski for creating this issue.