Hybrid collaborative recommendation algorithm based on WUDiff and RMF
A recommendation algorithm and algorithm technology, applied in computing, commerce, instruments, etc., can solve problems such as aggravated data sparseness, cold start, small storage capacity, and over-reliance on user-item matrix, etc., to achieve simple iteration, improve accuracy, and improve The effect of alleviating data sparsity
Inactive Publication Date: 2018-07-06
SOUTH CHINA NORMAL UNIVERSITY
View PDF4 Cites 10 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0008] With the rapid increase of a large number of new users and new items, problems such as data sparsity and cold start will be aggravated. Only the nearest neighbor similarity measure is used, especially when a large number of co-occurrence scores or co-occurrence labels are missing. The recommendation algorithm based on the nearest neighbor The recommendation quality and efficiency will decrease, and the method based on regularized matrix factorization has the advantages of simple iteration, fast convergence, and small storage capacity, but its disadvantage is that it relies too much on the user-item matrix and ignores the relationship between neighbors. Influence, especially with the large-scale increase in the number of users and resources, various problems caused by the curse of dimensionality will be unavoidable
Although the current method can improve the recommendation quality to a certain extent by utilizing the tag information, in the model combining the neighbor information and matrix factorization, the pairwise relationship between users-tag-items is not considered to find the neighbor users at the same time. , and use this kind of neighbor information to improve the RMF model to improve the shortcomings of traditional methods
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 moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0029] In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in combination with specific embodiments and with reference to the accompanying drawings.
[0030] The present invention uses the WUDiff algorithm to increase the consideration of the degree relationship between users and items, users and tags, and find a set of neighbors similar to the target user; then use the neighbor information to regularize the RMF model, thereby improving the accuracy of recommendation and effectively Solve the problem of scoring matrix sparsity. Such as figure 1 Shown is the resource allocation process of the three-part graph of users, items, and tags. The operation steps of the recommendation algorithm based on WUDiff are as follows:
[0031] Step A. Establish a tripartite graph. The specific resource-allocation process is that the target user u allocates the energy of its resource val...
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
Login to View More Abstract
The invention discloses a hybrid collaborative recommendation algorithm based on WUDiff and RMF. According to the algorithm, a WUDiff algorithm is utilized to abstract users, projects and tags into nodes in a tripartite graph; information hidden in the tripartite graph with a weight is utilized to mine association among deep potential similar users in a network based on scores and the tags, and aneighbor set similar to a target user is found; an RMF model is utilized to decompose a user-project score matrix into a user feature matrix and a project feature matrix, and data density is improvedthrough dimension reduction; and last, neighbor information obtained through the WUDiff algorithm is utilized to regularize the RMF model. Through the method, data can be processed from a global perspective, a main structure mode in original data can be easily discovered, and the method has the advantages of simple iteration, quick convergence and the like; and meanwhile, the data is understood from the perspective of the tripartite graph, a substance diffusion method is utilized to find the relations between every two of the users, the projects and the tags, the influence of information lossbrought by dimension reduction is avoided, prediction accuracy is effectively improved, and the data sparsity problem is relieved.
Description
technical field [0001] The present invention relates to an information recommendation algorithm, in particular to a hybrid collaborative recommendation algorithm based on WUDiff and RMF. Background technique [0002] In recent years, with the rapid development of the Internet and big data, the ever-increasing amount of network information has caused people to encounter the dilemma of "information overload". In order to solve this problem, the recommendation system came into being. It actively collects various information data of users (user registration information, user browsing logs, historical scoring records and project information, etc.), and mines users' hidden interests and behavior patterns. Analyze the results and changes in project information, adjust recommended content and service methods, and provide personalized recommendation services for users, such as Last.fm's music rating recommendation system, Amazon (Amazon) and JD's (JD) product recommendation systems, ...
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
Login to View More IPC IPC(8): G06Q30/06
CPCG06Q30/0631
Inventor陈洁敏李建国
OwnerSOUTH CHINA NORMAL UNIVERSITY