Purchases method and system
First Claim
1. An automated method for providing personalised purchase recommendations to a user, the method carried out on a mobile computing device programmed to carry out the method in combination with a server, said method comprising the steps of:
- generating a global probabilistic purchase model on the server of purchases in a general population over an entire range of users, the model providing recommendations for future purchases based on purchase transactions of the general population;
detecting on the mobile device, attributes of the user so as to identify personal characteristics relating to a behavior of the user;
calculating at least one correction factor, based on the detected attributes of the user; and
applying on the mobile device the correction factor to the global probabilistic purchase model to modify the recommendations of said model according to personal preferences of the user thereby generating a personalized recommendation list for future purchases for the user;
wherein the mobile device applies the correction factor to convert an output P(y|X) of the global probabilistic purchase model to an output {circumflex over (P)}(y|X) of a personalized probabilistic model for the user using a personalized prior {circumflex over (P)}(y) of the user wherein {circumflex over (P)}(y|X) is computed either according to Eq (1) which assumes equal global priors,
1 Assignment
0 Petitions
Accused Products
Abstract
In an automated method for providing personalised recommendations to a user, a global probabilistic purchase model based on prior interactions of a group of users of a system, is used in the generation of personalised recommendations for future purchases for a given user. Attributes of the given user are used to identify characteristics relating to the user'"'"'s personal purchasing, correction factors are calculated to update the output of the global probabilistic purchase model to personalise the recommendations for the user.
-
Citations
27 Claims
-
1. An automated method for providing personalised purchase recommendations to a user, the method carried out on a mobile computing device programmed to carry out the method in combination with a server, said method comprising the steps of:
-
generating a global probabilistic purchase model on the server of purchases in a general population over an entire range of users, the model providing recommendations for future purchases based on purchase transactions of the general population; detecting on the mobile device, attributes of the user so as to identify personal characteristics relating to a behavior of the user; calculating at least one correction factor, based on the detected attributes of the user; and applying on the mobile device the correction factor to the global probabilistic purchase model to modify the recommendations of said model according to personal preferences of the user thereby generating a personalized recommendation list for future purchases for the user; wherein the mobile device applies the correction factor to convert an output P(y|X) of the global probabilistic purchase model to an output {circumflex over (P)}(y|X) of a personalized probabilistic model for the user using a personalized prior {circumflex over (P)}(y) of the user wherein {circumflex over (P)}(y|X) is computed either according to Eq (1) which assumes equal global priors, - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14)
-
-
15. An automated system for providing personalised recommendations to a user, the system comprising:
-
a central unit comprising a General Recommendations Formulator capable of producing a global probabilistic purchase model from a Central Data Resource of purchases in a general population of an entire range of users, said model providing recommendations for future purchases as a function of purchase transactions of the general population; a module to detect attributes of the user so as to identify characteristics relating to the behavior of the user; means to calculate at least one correction factor, based on the attributes of the user; and means to apply the correction factor to the global probabilistic model to modify the recommendations of said model according to personal preferences of the user thereby generating a personalized recommendation list for future purchases for the user, wherein the means to apply the correction factor is configured to apply the correction factor to convert an output P(y|X) of the global probabilistic purchase model to an output {circumflex over (P)}(y|X) of a personalized probabilistic model for the user using a personalized prior {circumflex over (P)}(y) of the user, wherein {circumflex over (P)}(y|X) is computed either according to Eq (1) which assumes equal global priors, - View Dependent Claims (16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27)
-
Specification