Eureka AIR delivers breakthrough ideas for toughest innovation challenges, trusted by R&D personnel around the world.

Collaborative filtering method based on domain correlation self-adaption

A collaborative filtering and correlation technology, applied in the field of Internet recommendation system, can solve problems such as limited practicality, large amount of calculation, and difficulty in algorithm application, and achieve the effect of increasing calculation cost and strong practicability

Inactive Publication Date: 2016-12-14
HUAQIAO UNIVERSITY
View PDF1 Cites 5 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The debugging of parameters brings great difficulties to the application of the algorithm in practical problems
[0006] (2) Higher data requirements for auxiliary fields
These requirements on auxiliary domains greatly reduce the practicality of transfer learning methods
[0007] (3) For large-scale data, some migration learning algorithms require too much calculation
In practical applications, it usually involves processing massive users' ratings on items, and the excessive calculation required by the algorithm directly limits its practicability

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
  • Collaborative filtering method based on domain correlation self-adaption
  • Collaborative filtering method based on domain correlation self-adaption
  • Collaborative filtering method based on domain correlation self-adaption

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0035] The present invention will be further described below through specific embodiments.

[0036] refer to figure 1 , the method of the present invention mainly includes two steps: (1) Propose a new transfer learning model on the basis of the traditional matrix factorization model; (2) Use an iterative algorithm to solve the new model, and perform auxiliary domain and target domain correlation adaptive computing.

[0037] 1) Establish a new transfer learning model

[0038] Traditional model:

[0039] Assume that m users evaluate n items in the target field to form a scoring matrix T∈R m×n . Write Ω={(i,j)|the i-th user evaluates the j-th item}, then the set Ω represents the index set that has been evaluated in T. At present, a classic optimization model based on the nuclear norm regularization term is

[0040] m i n Z { 1 2 | ...

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

A collaborative filtering method based on domain correlation self-adaption comprises the following steps that 1,the difference of an auxiliary domain and a target domain serves as a regularization term and is introduced into a traditional model to obtain a new model (shown in the description), wherein T is a rating matrix having a partially deleted item in the target domain, Z and T have the same rating item; the equation shown in the description represents an index set representing the target domain, wherein the equation is shown in the description; | | . | | F represents a Frobenius norm, wherein the equation is shown in the description; | | . | | * represents a nuclear norm,| | Z | | *represents the sum of all singular values of a matrix Z; gamma is a regularization parameter, eta < (0, 1) represents the similarity of the auxiliary domain and the target domain; 2,a regular optimization solution Z* = Z of the new model is calculated by using a fixed-point iteration algorithm. The collaborative filtering method can be applied to an Internet recommendation system, introduces the system to a prediction model of the target domain through self-adaptive estimation of their correlation, accordingly achieves knowledge migration effectively and improves the recommendation precision of the target domain.

Description

technical field [0001] The invention relates to the field of data mining direction in information processing technology, in particular to a collaborative filtering method based on field correlation self-adaptation, which can be applied to an Internet recommendation system. Background technique [0002] With the increasing development of information and Internet of Things technology, the Internet meets the needs of users for information in the information age, brings convenience to people, and also brings a lot of information processing problems. When faced with massive amounts of information, users cannot obtain the part of information that is really useful to them, which makes the use of big data less efficient, and the so-called information overload problem arises. A very promising solution to the problem of information overload is the recommendation system. Recommender systems play an important role in many applications, for example, in e-commerce fields like Amazon, Tao...

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
IPC IPC(8): G06F17/30
CPCG06F16/9535
Inventor 王靖杜吉祥柳欣陈梦洁
Owner HUAQIAO UNIVERSITY
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
Eureka Blog
Learn More
PatSnap group products