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

Sequence recommendation method and device based on multi-scale interest dynamic hierarchy Transformer

A recommendation method and multi-scale technology, applied in the direction of instruments, data processing applications, business, etc., can solve the problem of low recommendation accuracy, achieve the effect of meeting individual needs and improving accuracy

Pending Publication Date: 2022-01-28
WUHAN UNIV
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] The present invention proposes a sequence recommendation method and device based on a multi-scale interest dynamic level Transformer, which is used to solve or at least partially solve the technical problem of low recommendation accuracy existing in the prior art

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
  • Sequence recommendation method and device based on multi-scale interest dynamic hierarchy Transformer
  • Sequence recommendation method and device based on multi-scale interest dynamic hierarchy Transformer
  • Sequence recommendation method and device based on multi-scale interest dynamic hierarchy Transformer

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0052] An embodiment of the present invention provides a sequence recommendation method based on a multi-scale interest dynamic hierarchy Transformer, including:

[0053] S1: For the input user history behavior sequence used to represent the interaction between the user and the item, learn the distributed representation of the user and the item;

[0054] S2: Based on the distributed representation of users and items, through the neighbor attention mechanism, neighbor blocks are merged layer by layer from bottom to top, and a hierarchical representation of multi-scale user interests of user historical behavior sequences is obtained;

[0055] S3: Model the relationship between the user interest at the next moment and the multi-scale user interest hierarchical representation of the user's historical behavior sequence through the self-attention mechanism of the Transformer layer, and obtain the user interest expression at the next moment;

[0056] S4: Concatenate the user interest...

Embodiment 2

[0108] Based on the same inventive concept, this embodiment provides a sequence recommendation device based on a multi-scale interest dynamic hierarchy Transformer, including:

[0109] The embedding module is used to learn the distributed representation of users and items for the input user historical behavior sequence to represent the interaction between users and items;

[0110] The dynamic hierarchical Transformer module is used for distributed representation based on users and items. Through the attention mechanism of neighbor blocks, neighbor blocks are merged layer by layer from bottom to top to obtain a hierarchical representation of multi-scale user interests of user historical behavior sequences;

[0111] The Transformer module is used to model the relationship between the user interest at the next moment and the multi-scale user interest hierarchical representation of the user's historical behavior sequence through the Transformer layer self-attention mechanism, and o...

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 a sequence recommendation method and device based on a multi-scale interest dynamic hierarchy Transformer, and the method comprises the steps of carrying out the fine-grained multi-scale modeling of the interest of a user through employing a sequence recommendation model, namely MIDHT (Multi-scale Interest Dynamic hierarchical Transformer); firstly, judging whether two adjacent articles need to be combined or not through a neighbor attention mechanism; 2, dynamically generating a block mask matrix by utilizing a judgment result in the previous step; 3, calculating implicit representation of a current layer according to the dynamic block mask matrix and a self-attention mechanism; and finally, deducing corresponding hierarchical structures from the dynamic block mask matrixes of all layers. The invention has the advantages of being high in operation efficiency and remarkable in individuation effect, and the final interest preference of the user can be well predicted. Therefore, the personalized recommendation performance is excellent.

Description

technical field [0001] The present invention relates to the technical field of e-commerce platform recommendation systems, in particular to a sequence recommendation method and device based on a multi-scale interest dynamic hierarchy Transformer. Background technique [0002] The main goal of the sequence recommendation task is to predict the items that the user will be interested in in the near future based on the user's historical behavior sequence. Different from traditional recommendation tasks, sequential recommendation considers the chronological order in which users interact with items, treats users' historical behavior data as sequences, and aims to capture the dynamic changes of users' interests. Different from time-sensitive recommendation tasks, sequential recommendation emphasizes the underlying variation patterns in the temporal order rather than the timestamps themselves. Different from content-based recommendation tasks, sequence recommendation only focuses o...

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): G06F16/9535G06F16/9536G06Q30/06G06Q50/00
CPCG06F16/9535G06F16/9536G06Q50/01G06Q30/0631
Inventor 胡瑞敏黄娜娜熊明福彭潇然潘昊李杰丁红卫
Owner WUHAN UNIV
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