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

User behavior prediction method and system based on deep walk and ensemble learning

A technology that integrates learning and prediction methods, applied in the field of machine recognition, can solve problems such as unguaranteed diversity, failure to effectively consider contextual semantic information, and overall classification performance degradation, so as to enhance prediction performance and reliability, and improve prediction reliability. reliability and prediction accuracy, and the effect of improving reliability and accuracy

Active Publication Date: 2020-09-22
湖南湖大金科科技发展有限公司
View PDF10 Cites 8 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] (1) When studying user behavior sequences, the contextual semantic information of each user behavior is not effectively considered, resulting in low learning ability and prediction accuracy of the trained model;
[0005] (2) When performing ensemble learning, most studies use random sampling to generate training subsets to construct several single classifiers, but their diversity cannot be guaranteed, which may lead to a decline in overall classification performance

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 behavior prediction method and system based on deep walk and ensemble learning
  • User behavior prediction method and system based on deep walk and ensemble learning
  • User behavior prediction method and system based on deep walk and ensemble learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0055] This embodiment proposes a user behavior prediction method based on deep walk and integrated learning.

[0056] In this embodiment, a commodity in the behavior sequence purchased by the user is regarded as a word, and all commodities are regarded as a document, so that some natural language processing techniques (NLP) can be used to train word vectors. On the other hand, in the scenario of user purchase behavior sequence, there is a large amount of graph structure information between data and data. These data information are very important. This embodiment applies DeepWalk technology to the purchase behavior network very well. in structure. Deep Walk (DeepWalk) technology uses random walk (Random Walk) technology to randomly walk the network nodes in the graph to form a behavior sequence. When the user's behavior sequence is regarded as a word, all behavior sequence documents are used Word2vec The algorithm model is used to pre-train word vectors, and on the basis of t...

Embodiment 2

[0077] In this embodiment, on the basis of the above-mentioned embodiment 1, a single model is further fused, such as image 3 shown. The fusion method of this embodiment first uses the maximum information coefficient (MIC) to measure the difference between each single learner respectively, and then expresses it in the form of a confusion matrix, thereby selecting two single learners with the largest difference for model Fusion for better generalization ability.

[0078] in:

[0079] 1. The maximum information coefficient (MIC) is used to measure the degree of correlation between two variables, whether it is a linear relationship or a nonlinear relationship. The calculation of maximum information coefficient (MIC) mainly utilizes mutual information (MI) and grid division method. Mutual information (MI) is used to measure the degree of association between two variables, given a variable set B={b 1 ,b 2 ,...,b n}, n is the number of samples, mutual information (MI) can be ...

Embodiment 3

[0104] In this embodiment, the background log of a certain bank shopping APP is taken as an example to test the method proposed in the above embodiment.

[0105] 1. Raw data set

[0106] The time span of the background log of the bank shopping app obtained is one month, mainly including more than 40,000 pieces of user consumption behavior data, each row corresponds to an operation record of the user, and is sorted according to the user's operation time. The relevant fields included in this data set are shown in Table 1.

[0107] Table 1 Basic information table of original data set

[0108]

[0109] Preprocess the original dataset.

[0110] 2. Construction of user portrait

[0111] Each line of records in the unprocessed data set is based on the user's operation behavior as the granularity, which is the information record of the user's single operation behavior, and this embodiment is to predict whether the user will buy a certain product, and each user needs to be more f...

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 behavior prediction method and system based on deep walk and ensemble learning. According to the method, preprocessing work is carried out on the problems of repetition,abnormality, redundancy and the like existing in an original data set, statistical information and activeness information capable of reflecting behavioral habits and preference degrees of consumers are extracted from the preprocessed data set to construct a user portrait for the user, then, random walk is carried out through a social network graph structure of commodities purchased by the user toobtain a new behavior sequence; and then, a Word2vec model is used to obtain the upper and lower information of each behavior of the user, and the upper and lower information is added into a machinelearning model for training and learning, so that the prediction reliability and prediction precision of the model are improved.

Description

technical field [0001] The invention relates to the technical field of machine recognition, in particular to a user behavior prediction method and system based on deep walk and integrated learning. Background technique [0002] With the rapid development of Internet technology and e-commerce, more and more people like to shop from the Internet to solve the problem of daily goods demand. Every day, thousands of users purchase products from e-commerce online shopping platforms. It is of great significance to use artificial intelligence algorithms to analyze user historical behaviors to determine whether the user buys the product. For example, researchers have found that by analyzing the historical shopping data of users on an e-commerce platform, they can mine out the characteristics of user preferences and behaviors, which has a great impact on personalized recommendations, user relationship management, and advertising costs. effect. In view of this, it is of great research...

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): G06F16/9536G06F16/955G06F40/30G06K9/62G06N3/04G06N3/08G06N20/10G06Q30/02
CPCG06F16/9536G06F16/955G06F40/30G06N3/084G06N20/10G06Q30/0202G06N3/045G06F18/2415Y02D10/00
Inventor 陈佐吴志良杨胜刚朱桑之谷浩然杨捷琳
Owner 湖南湖大金科科技发展有限公司
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