Recommendation system, method and non-transitory computer readable storage medium for storing thereof

Patent metadata

Field	Value
Publication number	US-9659302-B2
Application number	US-201314088534-A
Country	US
Kind code	B2
Filing date	Nov 25, 2013
Priority date	Oct 15, 2013
Publication date	May 23, 2017
Grant date	May 23, 2017

Abstract

A recommendation method includes providing an ontology database, in which the ontology database includes a plurality of entities, and the entities are arranged in an ontology hierarchy structure with N hierarchy levels; storing a plurality of j th level user data respectively corresponding to a plurality of users; generating a plurality of k th level user data according to the j th level user data respectively; clustering the k th level user data; and recommending the entities in the ontology database to the users according to a clustering result.

First claim

What is claimed is:

1. A recommendation method comprising:
providing an ontology database comprising an ontology hierarchy structure with N hierarchy levels, wherein N is an integer, and each of the hierarchy levels comprises at least one entity;
storing, through the ontology database, a plurality of j th user data respectively corresponding to a plurality of users, wherein each of the j th user data records at least one j th entity of the entities on a j th hierarchy level of the ontology hierarchy structure;
generating a plurality of k th user data corresponding to the users according to the j th user data respectively, wherein each of the k th user data records at least one k th entity of the entities on a k th hierarchy level of the ontology hierarchy structure;
clustering the k th user data; and
recommending the entities in the ontology database to the users according to the clustering result,
wherein the step of generating the k th user data corresponding to the users comprises:
calculating a sparsity of the j th user data; and
mapping the j th entity recorded in each of the j th user data to at least one of the k th entity recorded in each of the k th user data according to the sparsity of the j th user data;
wherein a first calculating value is equal to a product of a quantity of the users and a quantity of the entities on the j th hierarchy level, a second calculating value is equal to a quantity of a sum of the j th entity recorded by each of the j th user data divided by the first calculating value, and the sparsity of the j th user data is equal to 1 subtracted by the second calculating value.

2. The recommendation method as claimed in claim 1 , wherein the step of generating the k th user data corresponding to the users comprises:
mapping the j th entity recorded in each of the j th user data to at least one of the entities on the k th hierarchy level according to the ontology hierarchy structure, to serve as the k th entity recorded in each of the k th user data.

3. The recommendation method as claimed in claim 1 , wherein the step of mapping the j th entity recorded in each of the j th user data to at least one of the k th entity recorded in each of the k th user data comprises:
determining whether the sparsity of the j th user data is greater than a j th threshold; and
mapping the j th entity recorded in each of the j th user data to at least one of the k th entity recorded in each of the k th user data in a case that the sparsity of the j th user data is greater than the j th threshold.

4. The recommendation method as claimed in claim 3 , wherein the step of mapping the j th entity recorded in each of the j th user data to at least one of the k th entity recorded in each of the k th user data further comprises:
serving the j th user data as the k th user data in a case that the sparsity of the j th user data is not greater than the j th threshold.

5. The recommendation method as claimed in claim 1 , wherein the step of recommending the entities in the ontology database to the users according to the clustering result comprises:
searching a k th frequent entity from the k th entity recorded by each of the k th user data in a cluster;
searching a j th frequent entity from the entities on the j th hierarchy level according to the k th frequent entity; and
recommending the j th frequent entity to one of the users corresponding to the k th user data in the cluster.

6. The recommendation method as claimed in claim 5 , wherein the step of searching the j th frequent entity from the entities on the j th hierarchy level according to the k th frequent entity comprises:
determining whether the k th frequent entity is one of the entities on the j th hierarchy level;
mapping the k th frequent entity to corresponding entities of the entities on a k−1 th hierarchy level of the ontology hierarchy structure in a case that the k th frequent entity is not one of the entities on the j th hierarchy level; and
searching, through a frequent pattern mining algorithm, a k−1 th frequent entity from the corresponding entities on the k−1 th hierarchy level mapped by the k th frequent entity.

7. The recommendation method as claimed in claim 6 , wherein the step of searching the j th frequent entity from the entities on the j th hierarchy level according to the k th frequent entity further comprises:
serving the k th frequent entity as the j th frequent entity in a case that the k th frequent entity is one of the entities on the j th hierarchy level.

8. A recommendation system comprising:
a storage module configured to store an ontology database comprising an ontology hierarchy structure with N hierarchy levels, N is an integer, each of the hierarchy levels comprises at least one entity, the ontology database is configured to store a plurality of j th user data respectively corresponding to a plurality of users, and each of the j th data records at least one j th entity of the entities on a j th hierarchy level of the ontology hierarchy structure;
a converting module configured to generate a plurality of k th user data corresponding to the users according to the j data respectively, wherein each of the k th user data records at least one k th entity of the entities on a k th hierarchy level of the ontology hierarchy structure;
a clustering module configured to cluster the k th user data; and
a recommendation module configured to recommend the entities in the ontology database to the users according to the clustering result;
wherein the converting module is configured to calculate a sparsity of the j th user data, and map the j th entity recorded in each of the j th user data to at least one of the k th entity recorded in each of the k th user data according to the sparsity of the j th user data;
wherein a first calculating value is equal to a product of a quantity of the users and a quantity of the entities on the j th hierarchy level, a second calculating value is equal to a quantity of a sum of the j th entity recorded by each of the j th user data divided by the first calculating value, and the sparsity of the j th user data is equal to 1 subtracted by the second calculating value.

9. The recommendation system as claimed in claim 8 , wherein the converting module is configured to map the j th entity recorded in each of the j th user data to at least one of the entities on the k th hierarchy level according to the ontology hierarchy structure, to serve as the k th entity recorded in each of the k th user data.

10. The recommendation system as claimed in claim 8 , wherein the converting module is configured to determine whether the sparsity of the j th user data is greater than a j th threshold, and map the j th entity recorded in each of the j th user data to at least one of the k th entity recorded in each of the k th user data in a case that the sparsity of the j th user data is greater than the j th threshold.

11. The recommendation system as claimed in claim 10 , wherein the converting module is configured to serve the j th user data as the k th user data in a case that the sparsity of the j th user data is not greater than the j th threshold.

12. The recommendation system as claimed in claim 8 , wherein the recommendation module is configured to search a k th frequent entity from the k th entity recorded by each of the k th user data in a cluster, search a j th frequent entity from the entities on the j th hierarchy level according to the k th frequent entity, and recommend the j th frequent entity to one of the users corresponding to the k t

Assignees

• Inst Information Ind

Inventors

• Shih Meng-Jung
• Chen Ping-I
• Wang Tai-Chun
• Shae Zon-Yin

Classifications

• G06Q30/02 Primary

  Marketing; Price estimation or determination; Fundraising · CPC title