US 11,748,799 B2
Method, medium, and system for information recommendation
Sicong Chen, Beijing (CN); and Hai Yu, Beijing (CN)
Assigned to BEIJING JINGDONG SHANGKE INFORMATION TECHNOLOGY CO., LTD., Beijing (CN); and BEIJING JINGDONG CENTURY TRADING, CO., LTD., Beijing (CN)
Appl. No. 16/979,285
Filed by BEIJING JINGDONG SHANGKE INFORMATION TECHNOLOGY CO., LTD., Beijing (CN); and BEIJING JINGDONG CENTURY TRADING CO., LTD., Beijing (CN)
PCT Filed Mar. 11, 2019, PCT No. PCT/CN2019/077685
§ 371(c)(1), (2) Date Sep. 9, 2020,
PCT Pub. No. WO2019/174549, PCT Pub. Date Sep. 19, 2019.
Claims priority of application No. 201810199781.X (CN), filed on Mar. 12, 2018.
Prior Publication US 2020/0402137 A1, Dec. 24, 2020
Int. Cl. G06Q 30/00 (2023.01); G06Q 30/0601 (2023.01); G06Q 30/0201 (2023.01)
CPC G06Q 30/0631 (2013.01) [G06Q 30/0201 (2013.01); G06Q 30/0629 (2013.01); G06Q 30/0643 (2013.01)] 15 Claims
OG exemplary drawing
 
1. An information recommendation method, comprising:
identifying pairs of matched items according to historical data to determine cross-category relationships of the items;
constructing item matching scenes by using the cross-category relationships of the items and using an unsupervised learning method;
ranking the item matching scenes by using an item scene picture and a user scene picture, comprising:
obtaining combined features of a user and a scene by means of learning the item scene picture and the user scene picture using a strong learner;
obtain a coefficient of the combined features by using weak classifiers to score the obtained combined features; and
ranking the item matching scenes according to the coefficient of the combined features;
ranking items in the item matching scenes according to a feedback behavior on the items in the item matching scene from a user; and
recommending the ranked item matching scenes to the user.