tantivy icon indicating copy to clipboard operation
tantivy copied to clipboard

Add the notion of input schema mapping

Open fulmicoton opened this issue 6 years ago • 0 comments

Is your feature request related to a problem? Please describe. People may want to index a field more than once. Typically, someone may want to index a field with and without stemming, to make sure that the recall of stemming is there, but also that perfect term match get a scoring boost.

Right now this requires the user to duplicate its fields.

People may also want to index 2 fields in a single one for performance when scoring is not a problem. Right now we force them to concatenate the fields.

Describe the solution you'd like

Solution 1 In schema entries we could add an optional input_field parameter that decides which input field name it is populated from. By default, it would just be the same name as the schema field itself.

Solution 2 Entirely uncouple index schema and input schema, and add mapping information in a separate field in meta.json.

fulmicoton avatar Jan 25 '19 00:01 fulmicoton