Adaptive local low-rank matrix approximate modeling method based on recommendation system

A technology of low-rank matrix and modeling method, which is applied in the field of adaptive local low-rank matrix approximation method and approximate model, which can solve the problem of low performance of recommendation system

Pending Publication Date: 2020-09-01
BEIJING JIAOTONG UNIV
View PDF3 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0009] In order to improve the performance of the recommendation system and overcome the problem of low performance of the recommendation system under the background of big data, the present invention provides an adaptive local low-rank matrix approximation method and an approximate model (Adaptive LocalMatrix Appriximation, ALoMA) based on the recommendation system. Through the non-parametric unified Bayesian graphical model, while determining the scoring sub-matrix, determine the optimal rank of each sub-matrix, which significantly improves the recommendation performance in both scoring prediction and ranking prediction, and can provide a friendly Explained Recommended Results

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Adaptive local low-rank matrix approximate modeling method based on recommendation system
  • Adaptive local low-rank matrix approximate modeling method based on recommendation system
  • Adaptive local low-rank matrix approximate modeling method based on recommendation system

Examples

Experimental program
Comparison scheme
Effect test

no. 1 example

[0044] This embodiment provides a recommendation system-based adaptive local low-rank matrix approximate modeling method, the modeling method includes the following steps:

[0045] Step S1, constructing a scoring matrix.

[0046] The construction of scoring matrix described in this step is further, given a scoring matrix R=[R ij ] n×m , where the observable element R ij Records the ratings given by the i-th user to the j-th item.

[0047] In this step, usually, R is very sparse (the proportion of known values ​​is less than 1%), so the goal is to predict the set of unknown elements Π={(i,j):R ij missing}, and denote this set as R Π . Let X=[X ij ∈{0,1}] n×m Represents an indicator matrix, if R ij is the observed data, then X ij = 1; if R ij is missing data, then X ij =0.

[0048] Step S2, generating a scoring sub-matrix according to the scoring matrix.

[0049] In order to capture the local structure from the large-scale rating matrix R, users or items are divided...

no. 2 example

[0202] This embodiment provides a recommendation system-based adaptive local low-rank matrix approximation model (AdaptiveLocal Matrix Appriximation, ALoMA), figure 1 A schematic diagram of the frame of the model is shown. Such as figure 1 As shown, the model includes:

[0203] The score matrix, user-item sub-matrix and user-item feature representation matrix obtained by the method of the first embodiment above.

[0204] further, figure 2 A schematic diagram of the structure of the model is shown. Such as figure 2 As shown, the model includes:

[0205] Rating matrix R is a matrix composed of n users and m items, c i and d j represent the categories of the i-th user and the j-th item respectively, and the ij-th scoring entry (R ij ) is assigned to the cdth submatrix in; if the i-th user belongs to the c-th category (that is, c=c i ) and the j-th item belongs to the d-th category (ie, d=d j ), then the ijth scoring entry (R ij ) will be assigned to the cdth submat...

no. 3 example

[0208] This embodiment provides a recommendation method, which uses the recommendation system-based adaptive local low-rank matrix approximation model described in the second embodiment.

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention provides an adaptive local low-rank matrix approximate modeling method based on a recommendation system, a recommendation method and a recommendation system, which are used for solving the problem of low performance of a recommendation system under a big data background in the prior art. The invention discloses the self-adaptive local low-rank matrix approximate modeling method basedon a recommendation system. Firstly, a scoring matrix is constructed, then scoring sub-matrixes are generated, the rank of each sub-matrix is adaptively determined, local implicit representation in an optimal space is generated, an indication matrix is introduced for the scoring sub-matrixes, user/article bias is introduced, an observation scoring model is constructed, and thus an approximate model is constructed. According to the invention, local associated information in a subset of users or articles is captured; statistical capacity is allowed to be dynamically allocated between clusters,implicit representations of users or articles are adaptively mined from each sub-matrix through automatic association, the importance of potential features is determined through a missing mechanism, and recommendation performance is remarkably improved in score prediction and sorting prediction.

Description

technical field [0001] The invention belongs to the field of electronic commerce, and in particular relates to an adaptive local low-rank matrix approximation method and an approximation model based on a recommendation system. Background technique [0002] With the popularization of Internet technology, e-commerce has replaced physical stores and become an important economic activity in people's daily life and production activities. The recommendation system uses e-commerce websites to provide customers with product information and suggestions, to help users decide what products to buy, and to simulate salespeople to help customers complete the purchase process. Matrix approximation is one of the most effective collaborative filtering methods in recommender systems, which defines the recommendation problem as a prediction task for unobserved items in a sparse user-item rating matrix. The mainstream recommendation methods based on matrix approximation assume that the entire ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06Q30/06G06Q10/06
CPCG06Q30/0631G06Q10/067
Inventor 景丽萍刘华锋
Owner BEIJING JIAOTONG UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap