Personalized recommendation method and system, and computer-readable record medium

Patent metadata

Field	Value
Publication number	US-9665902-B2
Application number	US-201414563582-A
Country	US
Kind code	B2
Filing date	Dec 8, 2014
Priority date	Dec 8, 2014
Publication date	May 30, 2017
Grant date	May 30, 2017

Abstract

Provided is a method, system, and a computer-readable recording medium for providing a personalized recommendation of products. The method may include extracting product recommendations corresponding to a prescribed recommendation time period using two or more purchase cycle algorithms, the purchase cycle algorithms configured to calculate purchase cycles of products for a customer. The method may further include performing performance evaluation, using a processor, with respect to the product recommendations extracted using each of the purchase cycle algorithms, and recommending to the customer the product recommendation extracted from the purchase cycle algorithm having a highest ranking based on the performance evaluation.

First claim

What is claimed is:

1. A computer-implemented method of providing a personalized recommendation of products, comprising:
extracting product recommendations corresponding to a prescribed recommendation time period using a plurality of purchase cycle algorithms, the purchase cycle algorithms configured to calculate purchase cycles of products for a customer, wherein the purchase cycle is a prescribed period of time between each purchase of the same product;
performing a performance evaluation, using a processor, with respect to the product recommendations extracted using each of the purchase cycle algorithms, wherein performing the performance evaluation includes:
receiving a first data set of product purchase information and receiving a second data set of product purchase information, the second data set being a smaller data set than the first data set,
applying the first data set and the second data set to each of the purchase cycle algorithms and extracting a first product recommendation and a second product recommendation corresponding to the recommendation time period, and
obtaining hit rates of the first product recommendation and the second product recommendation based on a prescribed equation, wherein
hit rate (%)=[ M 1 ÷{N 1×( P 1 +P 2)÷2}]×100, and

wherein M 1 is a number of products in which the first product recommendation and the second product recommendation are matched, N 1 is a number of customers, P 1 is a number of first product recommendations, and P 2 is a number of second product recommendations; and

outputting, at a device of the customer, the product recommendation extracted from the purchase cycle algorithm having a highest ranking based on the performance evaluation,
wherein the plurality of purchase cycle algorithms includes a capacity purchase cycle algorithm that calculates a capacity purchase cycle according to capacity of the products, the capacity purchase cycle being based on an amount that the customer purchases during each purchase cycle, and
wherein the method is performed by a distributed processing system that includes a plurality of processors that operate in parallel and asynchronously.

2. The computer-implemented method of claim 1 , wherein the plurality of purchase cycle algorithms includes an average purchase cycle algorithm that calculates an average of periods of time between purchases of the same product.

3. The computer-implemented method of claim 2 , wherein the plurality of purchase cycle algorithms includes a standard deviation purchase cycle algorithm that applies a purchase cycle deviation to the average purchase cycle and calculates a standard deviation for the periods of time.

4. The computer-implemented method of claim 3 , wherein the standard deviation purchase cycle is obtained by a prescribed equation, wherein
standard deviation purchase cycle=average purchase cycle±purchase cycle deviation,

wherein the purchase cycle deviation is obtained from a square root of variance.

5. The computer-implemented method of claim 2 , wherein the average purchase cycle is obtained by a prescribed equation, wherein
average purchase cycle={(first period of time+second period of time + . . . + n -th period of time) ÷ n},

wherein the n-th period of time is a time between an n-th and a (n+1)-th purchase of the product.

6. The computer-implemented method of claim 1 , wherein the plurality of purchase cycle algorithms includes a most recent purchase cycle algorithm that calculates a last purchase cycle of the products.

7. The computer-implemented method of claim 1 , wherein the plurality of purchase cycle algorithms includes a minimum purchase cycle algorithm that obtains a minimum purchase cycle by selecting a shortest purchase cycle among the plurality of purchase cycle algorithms.

8. The computer-implemented method of claim 1 , wherein the capacity purchase cycle is obtained by a prescribed equation, where
capacity purchase cycle=[{(first period of time+second period of time+ . . . + n -th period of time)÷total capacity of purchase product before n-th purchase cycle}×capacity of purchase product of last purchase date].

9. The computer-implemented method of claim 1 , wherein the first data set is a learning data and the second data set is a verifying data, wherein the learning data and the verifying data are classified by a ratio of an amount of data or a time period.

10. The method of claim 1 , wherein extracting the product recommendations includes generating a list of products using the purchase cycle algorithms, and wherein the method further comprises:
ranking the purchase cycle algorithms with respect to the customer based on the performance evaluation;
generating an order list of products based on the ranking the purchase cycle algorithms; and
generating an output that includes at least a subset of the ordered lists of products, wherein the output is presented by the device of the customer.

11. A system for providing a personalized recommendation of products, comprising:
a data management server configured to store purchase history information of a customer;
a distributed processing system configured to determine the product recommendation for the customer using data provided from the data management server, wherein the distributed processing system includes a plurality of processors that operate in parallel and asynchronously; and
an output device configured to provide the product recommendation determined by the distributed processing system to the customer,
wherein the distributed processing system is further configured to:
extract product recommendations corresponding to a prescribed recommendation time period using a plurality of purchase cycle algorithms, the purchase cycle algorithms configured to calculate purchase cycles of products for a customer, wherein the purchase cycle is a prescribed period of time between each purchase of the same product;
perform a performance evaluation with respect to the product recommendations extracted using each of the purchase cycle algorithms, wherein the distributed processing system, when performing the performance evaluation, is further configured to:
determine a first data set of product purchase information and a second data set of product purchase information, the second data set being a smaller data set than the first data set,
apply the first data set and the second data set to each of the purchase cycle algorithms and extracting a first product recommendation and a second product recommendation corresponding to the recommendation time period, and
obtain hit rates of the first product recommendation and the second product recommendation based on a prescribed equation, wherein
hit rate(%)=[ M 1 ÷{N 1×( P 1 +P 2)÷2}]×100, and

wherein M 1 is a number of products in which the first product recommendation and the second product recommendation are matched, N 1 is a number of customers, P 1 is a number of first product recommendations, and P 2 is a number of second product recommendations; and
transfer, for output at the output device, the product recommendation extracted from the purchase cycle algorithm having a highest ranking based on the performance evaluation, and

wherein the plurality of purchase cycle algorithms includes a capacity purchase cycle algorithm that calculates a capacity purchase cycle according to capacity of the products, the capacity purchase cycle being based on an amount that the customer purchases during each purchase cycle.

12. The personalized recommendation system of claim 11 , wherein the output device provides the product recommendation to customers connected to the data management server, or to a te

Assignees

• Lg Cns Co Ltd

Inventors

• Lee Hwan Muk
• Lee Dae Sik
• Kim Byung Keug
• Kim Soo Min
• Park Min Yong
• Song Han Yi
• Lee Dong Yoon
• Lee Yoon Jae

Classifications

• G06Q30/0255

  based on user history · CPC title

• G06Q30/02

  Marketing; Price estimation or determination; Fundraising · CPC title

• G06Q30/0282

  Rating or review of business operators or products · CPC title

• G06Q30/0201

  Market modelling; Market analysis; Collecting market data · CPC title

• G06F7/00

  Methods or arrangements for processing data by operating upon the order or content of the data handled (logic circuits H03K19/00) · CPC title