embed_and_search ranks with the Search Tuning saved for your index. Each parameter below changes one setting for that call only. The saved settings do not change, and the hosted search page keeps using them.
01 · The parameters
Add them to an embed_and_search call. The default is what an index uses until you change it in the panel.
| Parameter | Setting in the panel | Values | Default |
|---|---|---|---|
lexical_weight | Semantic ↔ Lexical Balance | From 0.10 (meaning leads) to 1.00 (only the words decide) | 0.15 |
fw_titlefw_descriptionfw_urifw_textfw_text_t | Field Weights: Title, Description, URI, Text, LD Text | A number, 0.01 or more; only how they compare matters | 0.20 0.05 0.01 0.01 0.01 |
mm | Minimum Match | flexible, balanced, strict, or your own rule such as 75% or 2<90% 5<75% | flexible |
search_mode | Search Mode | union, keywords_required, meaning_required, intersection | union |
vector_topk | Vector Candidate Pool | From 10 to 1000 | 500 |
quality_boost | Content Quality Boost | From 0 (off) to 1 | 0 |
fresh_bias | Fresh Results Bias, the Fresh switch of the hosted page | 1 turns it on for the call, 0 turns it off. On, newer documents rank higher and undated ones are left out, with the index's Fresh Results Bias Weight. | off |
freshness_boost | not in the panel | The window of fresh=yes, in days | 7 |
A number setting sent with something that is not a number, or a search_mode that is not one of the four above, is ignored, and the saved setting applies. Lexical Only is a setting of the panel and cannot be chosen per call.
02 · Example
curl -s -G https://api.opensolr.com/solr_manager/api/embed_and_search \ --data-urlencode "email=YOUR_EMAIL" \ --data-urlencode "api_key=YOUR_API_KEY" \ --data-urlencode "index_name=YOUR_INDEX" \ --data-urlencode "q=thermal camera for walls" \ --data-urlencode "fresh=no" \ --data-urlencode "search_mode=keywords_required" \ --data-urlencode "mm=strict" \ --data-urlencode "fw_title=0.5"
The debug part of the answer shows the search mode and the vector pool the call used, and params shows the rest.
03 · Related
The call itself, its answer and its errors: embed_and_search.
What each setting does, in plain words: Search Tuning.