Difference between revisions of "Server-side product indexing and search - Technical Specs"

From Open Food Facts EN
Jump to navigation Jump to search
(Created page with "== Summary == This page contains the technical specifications for the Server-side product indexing and search task of the Project:Personalized_Search project. It li...")
 
 
Line 16: Line 16:
  
 
The search API will return a large number of products (e.g. 100 products) that match the query and order them with this pre-computed key.
 
The search API will return a large number of products (e.g. 100 products) that match the query and order them with this pre-computed key.
 +
 +
The Products::compute_sort_keys() function computes the popularity_key key.
  
 
== Search API ==
 
== Search API ==
  
 
=== Search by category ===
 
=== Search by category ===
 +
 +
Currently working:
 +
* https://fr.openfoodfacts.org/cgi/search.pl?action=process&tagtype_0=categories&tag_contains_0=contains&tag_0=en%3Acookies&sort_by=popularity_key&page_size=20&action=display
 +
 +
Possible change of syntax for filters:
 +
* categories=some category name, or categories=en:some-category-id
  
 
=== Search by keyword ===
 
=== Search by keyword ===
  
 +
Currently working:
 +
* https://fr.openfoodfacts.org/cgi/search.pl?action=process&search_terms=soda&sort_by=popularity_key&page_size=20&action=display&json=1
  
 
[[Category:Project:Personalized_Search]]
 
[[Category:Project:Personalized_Search]]

Latest revision as of 16:02, 29 June 2020

Summary

This page contains the technical specifications for the Server-side product indexing and search task of the Project:Personalized_Search project.

It lists the options we are considering and the one we choose.

Server-side sort order pre-computation

The current API returns results sorted by last modified date.

For the Personalized Search project, we will pre-compute for each product a query independent sort key based on:

  • data completeness: how much key data we have for the product
  • data quality: how confident we are that the data is correct and up-to-date
  • product quality: how high is the overall quality of the product (e.g. nutritional quality)
  • product popularity: how popular the product is

The search API will return a large number of products (e.g. 100 products) that match the query and order them with this pre-computed key.

The Products::compute_sort_keys() function computes the popularity_key key.

Search API

Search by category

Currently working:

Possible change of syntax for filters:

  • categories=some category name, or categories=en:some-category-id

Search by keyword

Currently working: