System, apparatus, and method for recommending items

• US 9,298,809 B2
• Filed: 03/15/2013
• Issued: 03/29/2016
• Est. Priority Date: 03/30/2012
• Status: Active Grant

• Current Assignee: Sony Corporation (Sony Group Corp.)
• Original Assignee: Sony Corporation (Sony Group Corp.)
• Inventors: Kemp, Thomas, Kamimaeda, Naoki, Miyahara, Masanori
• Primary Examiner(s): Ruiz, Angelica
• Application Number:
  US13/837,508
• Publication Number:
  US 20130262478A1
• Time in Patent Office: 1,110 Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/3334 Selection or weighting of terms from queries, including natural language queries
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising

Abstract

Recommending one or more items among a plurality of items to a first user by generating a weighted data set by applying a weight factor to each of a plurality of user event data sets each derived from a previous user event of a second user of the recommendation system. Each of the user event data sets includes a user identifier of the second user, an item identifier of an item involved in the previous user event, and time information regarding a time of the previous user event. The weight factor is derived from the time information of the corresponding previous user event. One or more recommended items are determined based on an algorithm using the weighted data set.

Claims

1. A method performed by an information processing apparatus for recommending one or more items among a plurality of items to a first user of a recommendation system, the method comprising:

deriving, using circuitry of the information processing apparatus, a first weight factor for a plurality of user event data sets from an order of previous user events of a second user;

generating, using the circuitry, a weighted data set by applying the first weight factor to each of the plurality of user event data sets each derived from a previous user event of the second user of the recommendation system, wherein

each of the user event data sets includes a user identifier of the second user, an item identifier of an item involved in the previous user event of the second user, and time information regarding a time of the previous user event of the second user, and

the first weight factor is derived from the time information of the previous user event of the second user; and

determining, using the circuitry, the one or more recommended items based on an algorithm using the weighted data set,

wherein the order of previous user events of the second user is ordered according to most recent user event of the second user to least recent user event of the second user, respective weights for the ordered previous user events of the second user descending from highest weight for the most recent user event to lowest weight for the least recent user event.

2. The method according to claim 1, further comprising:

deriving, using the circuitry, the first weight factor of the user event data sets from a time difference between a current time and the time of the previous user event of the user event data set.

3. The method according to claim 2, wherein said deriving includes deriving the first weight factor of the user event data sets from an exponential function of the time difference between the current time and the time of the previous user event of the user event data set.

4. The method according to claim 2, wherein said deriving includes deriving the first weight factor of the user event data sets from a linear function of the time difference between the current time and the time of the previous user event of the user event data set.

5. The method according to claim 1, further comprising:

determining, using the circuitry, similarities between at least one of users of the recommendation system and between items among the plurality of items; and

determining, using the circuitry, recommendation values for the one or more items based on the determined similarities and based on the weighted data set.

6. The method according to claim 5, wherein said determining similarities includes determining similarities based on the weighted data set.

7. The method according to claim 1, wherein the user events are purchase events.

8. The method according to claim 1, further comprising:

generating, using the circuitry, a matrix based on the plurality of user event data sets; and

applying, using the circuitry, a second weight factor to one or more elements of the matrix.

9. An information recommendation apparatus comprising:

circuitry configured to

derive a weight factor for a plurality of user event data sets from an order of previous user events of a second user,

generate a weighted data set by applying the weight factor to each of the plurality of user event data sets, each derived from a previous user event of the second user of the apparatus, wherein

each of the user event data sets includes a user identifier of the second user, an item identifier of an item involved in the previous user event of the second user, and time information regarding a time of the previous user event of the second user, and

the weight factor is derived from the time information of the previous user event of the second user;

determine one or more recommended items based on an algorithm using the weighted data set; and

store the plurality of user event data sets in a memory,

wherein the order of previous user events of the second user is ordered according to most recent user event of the second user to least recent user event of the second user, a weight assigned to the most recent user event of the second user being higher than any other weight assigned to other ordered previous user events, including said least recent user event of the second user.

10. The information recommendation apparatus according to claim 9, wherein the circuitry is configured to store the weighted data set in the memory.

11. A non-transitory computer-readable storage medium storing computer-readable instructions that, when executed by a computer of a recommendation apparatus, cause the recommendation apparatus to perform a method comprising:

deriving a weight factor for a plurality of user event data sets from an order of previous user events of a second user;

generating a weighted data set by applying the weight factor to each of the plurality of user event data sets, each derived from a previous user event of the second user of a recommendation system, wherein

each of the user event data sets includes a user identifier of the second user, an item identifier of an item involved in the previous user event of the second user, and time information regarding a time of the previous user event of the second user, and

the weight factor is derived from the time information of the previous user event of the second user; and

determining one or more recommended items based on an algorithm using the weighted data set,

wherein the order of previous user events of the second user is ordered according to most recent user event of the second user to least recent user event of the second user.