Nearest-neighbor collaborative filtering method based on item feature augmentation (IFA) of product

A collaborative filtering and nearest neighbor technology, applied in special data processing applications, instruments, electrical digital data processing, etc., can solve the problems of reducing the accuracy of recommendation systems, calculating similarity measurement errors, and inaccurate similarity calculations, etc.

CN106055715AActive Publication Date: 2016-10-26HEFEI UNIV OF TECH
1 Cites 5 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Publication Date
2016-10-26

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
  • Figure 3
    Figure 3
Patent Text Reader

Abstract

The invention discloses a nearest-neighbor collaborative filtering method based on item feature augmentation (IFA) of a product. The method comprises the following steps of: (1) representing scoring data of the product with a two-dimensional table; (2) calculating an item similarity of a poor information user based on a naive bayes classifier; (3) calculating the item similarity of a rich information user; (4) fusing the item similarity of the poor information user and the item similarity of the rich information user; and (5) giving out a scoring predictive value set of products which are not scored in combination with an item-based collaborative filtering algorithm, and taking first N items as a final recommendation result. According to the method, the true situation of user evaluation can be effectively reflected and the individualized recommendation precision is improved.
Need to check novelty before this filing date? Find Prior Art

Description

technical field

[0001] The invention belongs to the field of electronic commerce, and specifically relates to a nearest neighbor collaborative filtering method (Item Feature Augmentation, IFA) based on product item feature expansion. Background technique

[0002] With the rapid development of e-commerce, how to effectively increase the user purchase rate has become the main consideration of major e-commerce companies. Collaborative filtering technology, as one of the earliest and most successful technologies for personalized recommendation applications, can be well based on items or users. The similarity between them provides technical support for the purchase decision of the enterprise. The collaborative filtering recommendation technology based on the nearest neighbor of the item has become the most popular algorithm based on the nearest neighbor collaborative filtering algorithm because of its strong scalability and good explainability. Shopping in Amazon It has seen prac...

Examples

Embodiment Construction

[0064] like figure 1 As shown, a nearest neighbor collaborative filtering method (Item FeatureAugmentation, IFA) based on product item feature expansion is carried out as follows:

[0065] Step 1, using a two-dimensional table T={U, I, S} to represent the rating data of the product;

[0066] In the two-dimensional table T, U={U 1 ..., U u ,...,U u} represents the set of users, I={I 1 ..., I i ,...,I |i|} means product set, S={S (1) ..., S (s) ,...,S (s)} represents the collection of user ratings on the product;

[0067] In the user set U, |u| is the total number of users, U u Indicates the uth user; 1≤u≤|u|; in the product set I, |i| is the total number of products, I i Indicates the i-th product; 1≤i≤|i|; in the scoring set S, S (s) Indicates the sth rating in the rating set, and S (1) (s) (s) ;

[0068] Let the u-th user U u For the i-th product I i rated as S u,i , and S u,i ∈ S;

[0069] like figure 2 As shown, for any ith product I i and the jth prod...