# US 8,572,017 B2 · User-based collaborative filtering recommendation system and method for amending similarity using information entropy

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 8,572,017 B2
- Document ID: US8572017B2
- Status: Granted
- Kind: B2
- Application number: 12/692,575
- Filed: 2010-01-22
- Published: 2010-07-22
- Granted: 2013-10-29
- Earliest claimed: 2009-01-22
- Assignees: Sungkyunkwan University Foundation for Corporate Collaboration
- Applicants: Kwang Seok Hong; Hyeong Joon Kwon
- Inventors: Kwang Seok Hong; Hyeong Joon Kwon
- Web page: https://patentgrep.com/patent/US8572017B2

## CPC

- G06N 5/04: Inference or reasoning models
- G06F 16/954: Navigation, e.g. using categorised browsing
- G06Q 30/02: Marketing; Price estimation or determination; Fundraising
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR

## Abstract

The present invention relates to a user-based collaborative filtering recommendation method and system and, more specifically, relates to a user-based collaborative filtering recommendation method and system which amends the similarity between the recommend probable user and the other user to improve the accuracy in the prediction of preference by using the relation between the user preference information entropies of the recommend probable user and the other user.

## Claims

### Claim 1

A user-based collaborative filtering recommendation method comprising:
a first step of calculating the similarity between a user A and another user B;
a second step of amending the calculated similarity by using each user preference information entropy of the user A and the user B;
a third step of forming a nearest neighbor list of the user A based on the amended similarity; and
a fourth step of predicting the preference of the user A by using the preference of the users in the formed nearest neighbor list,
wherein the second step of amending the similarity comprises the steps of:
obtaining the information entropy H(A) of the user A and the information entropy H(B) of the user B;
obtaining a difference value between the information entropy H(A) of the user A and the information entropy H(B) of the user B;
obtaining a reciprocal number of the value obtained by adding 1 to the difference value between the H(A) and H(B); and
obtaining the value obtained by multiplying the similarity between the user B and the user A calculated in the similarity calculating step by the reciprocal number of the value obtained by adding 1 to the difference value between H(A) and H(B) as amended similarity.

### Claim 2 (depends on 1)

The user-based collaborative filtering recommendation method of claim 1, further comprising a fifth step of letting the user A input actual preference for an item for which the user A does not express his preference before and comparing the input actual preference with the preference predicted in the fourth step.

### Claim 3 (depends on 1)

The user-based collaborative filtering recommendation method of claim 1, wherein the fourth step is performed after the first, the second and the third steps are performed for all users.

### Claim 4

A user-based collaborative filtering recommendation method comprising:
a first step of calculating the similarity between a user A and another user B;
a second step of amending the calculated similarity by using each user preference information entropy of the user A and the user B;
a third step of forming a nearest neighbor list of the user A based on the amended similarity; and
a fourth step of predicting the preference of the user A by using the preference of the users in the formed nearest neighbor list,
wherein the second step of amending the similarity is to amend the similarity by using the relation between the user preference information entropy of the user B and the user preference information entropy of the user A as a weight value,
wherein a weight value that is multiplied by the similarity calculated in the first step is more than 0 and less than 1.

### Claim 5 (depends on 4)

The user-based collaborative filtering recommendation method of claim 4, further comprising a fifth step of letting the user A input actual preference for an item for which the user A does not express his preference before and comparing the input actual preference with the preference predicted in the fourth step.

### Claim 6 (depends on 4)

The user-based collaborative filtering recommendation method of claim 4, wherein the fourth step is performed after the first, the second and the third steps are performed for all users.