Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

User recommendation method

A recommendation method and user technology, applied in neural learning methods, data processing applications, special data processing applications, etc., can solve problems such as data sparseness, achieve the effect of improving robustness, solving data sparse problems, and enriching information

Active Publication Date: 2018-04-20
NORTHEAST NORMAL UNIVERSITY
View PDF9 Cites 10 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a user recommendation method, which can solve the problem of data sparseness in traditional recommendation systems, and can more accurately predict user preferences and recommend items to users

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
  • User recommendation method
  • User recommendation method
  • User recommendation method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0029] The TDAE model is a model that integrates the user's rating information and the user's explicit trust relationship in the social network to predict and score the user's preference. figure 1 The flow chart of the user recommendation method provided in this application. Such as figure 1 shown, including the following steps:

[0030] Step S11: Determine the input data from the encoder.

[0031] Specifically, this application uses x∈R in the autoencoder N The vector of is used as input data. where x represents the input vector of the autoencoder, R N Indicates that the dimension of the vector x is N.

[0032] Step S12: Map the input data to the hidden layer of the autoencoder through an activation function. Among them, the specific representation of the hidden layer is as follows:

[0033] y=ρ(W T x+b)

[0034] Among them, y represents the result of the hidden layer, ρ represents the activation function, W represents the weight matrix of N×H dimensions, and W T Re...

Embodiment 2

[0065] The TDAE++ model is based on the TDAE model, which extracts the implicit trust relationship between users through the similarity measurement method, and then combines the explicit trust information and implicit trust information with scoring data to improve the quality of predictive scoring . Based on the first embodiment above, training model parameters on the data set further includes step S143: extracting implicit information from the data set through a similarity measurement method. On this basis, it is necessary to combine scoring data, explicit trust information and implicit information when predicting ratings.

[0066] Such as image 3 As shown, it specifically includes the following sub-steps:

[0067] Step S311: Measure the implicit trust relationship between users.

[0068] As an embodiment, this application uses similarity to measure the implicit trust relationship between users, which is specifically expressed as follows:

[0069]

[0070] Among them,...

Embodiment 3

[0083] This application selects PMF and DAE models as comparison models, and selects mean absolute error and root mean square error as evaluation criteria to measure the effect of the model.

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 discloses a user recommendation method. The method specifically comprises the following steps of: determining input data of an auto-encoder; mapping the input data to an implicit layer of the auto-encoder through an activation function; mapping the implicit layer into a one-dimensional reconstructed vector; and training model parameters in a data set to obtain a predicted score of auser, wherein the step of training model parameters is carried out through minimizing a reconstruction error. The step of training the model parameters comprises the following steps of: carrying out sparseness on the input data of the auto-encoder; and carrying out training by combining score data of the data set and explicit trust information. The step of carrying out sparseness on the input dataof the auto-encoder comprises the following sub-steps of: extracting an input vector from a user-project score matrix in the data set; setting a missing value in the input value to be zero; adding masking noise in the auto-encoder; and before counter-propagation, setting an error of the missing value in the input vector to be zero. The method is capable of achieving the effect of improving the recommendation correctness.

Description

technical field [0001] The present application relates to the fields of information filtering and data mining, and in particular to a user recommendation method. Background technique [0002] Recommender systems are widely used in various electronic service systems, such as e-commerce and social networks. The recommendation system can recommend new items to users based on their historical ratings. Two commonly used approaches in recommender systems are content-based filtering and collaborative filtering. Content-based filtering is to recommend new items to users by analyzing the items that users liked in the past. Collaborative filtering is to use a set of known user preferences to recommend items or score predictions to users with unknown preferences. Traditional collaborative filtering models users only through their rating data, however, the sparsity of ratings will seriously reduce the effectiveness of recommendation systems. [0003] Currently, content-based filteri...

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): G06F17/30G06N3/04G06N3/08G06Q30/02G06Q30/06G06Q50/00
CPCG06F16/337G06N3/084G06Q30/0201G06Q30/0202G06Q30/0631G06Q50/01G06N3/048
Inventor 张邦佐王美琪武志远孙小新冯国忠
Owner NORTHEAST NORMAL UNIVERSITY
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com