Intelligent user portrait method based on small data input

A technology for small data and users, applied in special data processing applications, neural learning methods, electrical digital data processing, etc. , weaken the effect of the Matthew effect

Active Publication Date: 2019-11-15
HUA DATA TECH (SHANGHAI) CO LTD
View PDF10 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Its disclosed technical solution combines the Internet of Things, big data, and intelligent portrait technology, and can give professional medical advice to medical examiners. However, it collects various information of medical examiners and uses a large amount of user data to construct user portraits, which cannot solve small problems. Data cold start problem

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
  • Intelligent user portrait method based on small data input

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0024] The present invention is further described below:

[0025] The invention discloses an intelligent user portrait method based on small data input, which includes:

[0026] Create user basic information model, behavior latitude model, input user corresponding data, generate basic information behavior data;

[0027] Carry out deep learning on basic information behavior data to obtain high-level information data of user behavior;

[0028] Through the feed-forward neural network, the high-order information data of user behavior is mapped to the hidden internal drive model to obtain the hidden internal drive data;

[0029] Create user cross-domain behavior model data;

[0030] Match implicit drive data with user cross-domain behavior model data to generate user portraits.

[0031] Specifically, the user basic information is input into the user basic information model, and the user behavior data is input into the behavior latitude model.

[0032] Further, the basic informa...

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 an intelligent user portrait method based on small data input, and the method comprises the steps: building a user basic information model and a behavior latitude model, inputting user corresponding data, and generating basic information behavior data; performing deep learning on the basic information behavior data to obtain user behavior high-order information data; mapping the user behavior high-order information data into a recessive internal drive model through a feedforward neural network to obtain recessive internal drive data; creating user cross-domain behaviormodel data; and matching the implicit internal drive force data with the user cross-domain behavior model data to generate a user portrait. According to the method, after earlier-stage data collectionand processing are completed, the implicit internal drive force data is matched with the user cross-domain behavior model data, and the user portrait is generated. A hidden internal drive force (BFI)technology is adopted, so that the dependence on data is greatly reduced, and small data cold start is supported; output dimensions are rich, and cross-domain prediction is supported; mitigating Maieffect.

Description

technical field [0001] The invention relates to a user portrait method, in particular to an intelligent user portrait method based on small data input. Background technique [0002] The user profiling technology in the prior art is a technology that tags users by mining user behavior data, and has many applications in interest mining, advertisement recommendation, anomaly detection, and the like. [0003] The traditional user portrait technology is mainly based on the collaborative filtering algorithm, which requires a large amount of multi-dimensional behavior data in the training process. When the user behavior data is sparse or the dimensions are sparse, the accuracy of the description of the user portrait is poor, not only It will affect commercial use, and at the same time, it will bring a lot of misleading. [0004] In addition, the traditional user portrait technology has a strong Matthew effect, and its performance on long-tail mining is very low. Therefore, suppor...

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): G06F16/9536G06N3/04G06N3/08
CPCG06N3/04G06N3/08G06F16/9536
Inventor 徐清
Owner HUA DATA TECH (SHANGHAI) CO LTD
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
Try Eureka
PatSnap group products