US 11,816,721 B2
Recommendation apparatus and method
Martyna Ksyta, Hatfield (GB); Jakub Smajek, Hatfield (GB); Michal Źelechowski, Hatfield (GB); Jose Jimenez, Hatfield (GB); Przemyslaw Pastuszka, Hatfield (GB); and Maciej Mnich, Hatfield (GB)
Assigned to OCADO INNOVATION LIMITED, Hatfield (GB)
Appl. No. 16/964,563
Filed by OCADO INNOVATION LIMITED, Hatfield (GB)
PCT Filed Jan. 24, 2019, PCT No. PCT/EP2019/051704
§ 371(c)(1), (2) Date Jul. 23, 2020,
PCT Pub. No. WO2019/145395, PCT Pub. Date Aug. 1, 2019.
Claims priority of application No. 1801228 (GB), filed on Jan. 25, 2018; and application No. 1803405 (GB), filed on Mar. 2, 2018.
Prior Publication US 2021/0035188 A1, Feb. 4, 2021
Int. Cl. G06Q 30/00 (2023.01); G06Q 30/0601 (2023.01); G06N 20/00 (2019.01); G06F 17/18 (2006.01)
CPC G06Q 30/0631 (2013.01) [G06F 17/18 (2013.01); G06N 20/00 (2019.01)] 20 Claims
OG exemplary drawing
 
1. A recommendation computer arranged to communicate with a product information database, a product similarity database, a customer purchase history database and a customer order database, the recommendation computer comprising:
program code that, when executed, causes the recommendation computer to:
determine at least one similarity between information about a product stored in the product information database and information about a product stored in a customer's virtual basket and, if there is a determined similarity, store the at least one determined similarity in the product similarity database, wherein the similarity is based on at least one of a product category, a product name, product ingredients, and products in a recipe;
train a model for calculating a probability that a product should be included in a customer's current order based on information about products absent from the customer's virtual basket and previously purchased by at least the customer as stored in the customer purchase history database and information about at least one determined similarity relating to the previously purchased products absent from the customer's virtual basket including at least one of: if the customer has searched for a similar product and if the customer has already browsed for a particular/similar product but has not added it to the current order, as stored in the product similarity database;
generate information regarding at least one product that is absent from the customer's virtual basket and similar to at least one product to be purchased by the customer based on information about the current order of the customer stored in the customer order database and information about determined similarities as stored in the product similarity database;
provide the at least one product determined to be similar to at least one product included in the customer's current order and that is not currently stored in the customer's virtual basket as an input to the trained model, and, for each at least one similar product, calculate a probability that the at least one similar product should be included in the current order, the probability being calculated as a function of at least preferred products identified by the customer and a purchase history of the customer;
recommend to a computing device of the customer, product or products which should be included in the order based on the probability or probabilities output by the trained model; and
retrain the model based on recommended products selected for purchase by the customer.