Real-time hybrid recommendation method and system based on Labda architecture

A mixed recommendation and content recommendation technology, applied in marketing, advertising, instruments, etc., can solve the problem that components are not up to the task, achieve high throughput, ensure relevance, enhance real-time and accuracy

Pending Publication Date: 2020-05-05
FUDAN UNIV
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Although there are various mature open source components that can handle massive data and streaming data, a single component is not up to the task

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
  • Real-time hybrid recommendation method and system based on Labda architecture
  • Real-time hybrid recommendation method and system based on Labda architecture
  • Real-time hybrid recommendation method and system based on Labda architecture

Examples

Experimental program
Comparison scheme
Effect test

experiment example 1

[0058] Experimental Example 1: Determination of parameter β in mixed recommendation, see Figure 10 shown. When β=0, the content recommendation algorithm based on relative word frequency works alone, and when β=1.0, the collaborative filtering based on incremental update works alone. It can be seen that when 0≤β≤0.4 and 0.5≤β≤0.7, as the weight of collaborative filtering based on incremental update increases in hybrid recommendation, F 1 The value is also increasing and reaches a maximum at β=0.7, therefore, in all subsequent experiments, β=0.7 is taken.

experiment example 2

[0059] Experimental Example 2: Analysis of Recommendation Effects of Different Recommendation Algorithms

[0060] Figure 11 It shows the performance of each algorithm in terms of accuracy indicators under different recommendation list lengths. When the number of news candidates recommended to users increases, the value of accuracy first increases and then decreases. When the length of the recommendation list is 15, each recommendation method achieves the best recommendation effect.

experiment example 3

[0061] Experimental example 3: Recommendation system performance evaluation, see Figure 12 shown.

[0062] When the number of cluster computing nodes remains unchanged, the traffic peak value continues to increase, and the computing time increases gradually. This shows that as the number of visitors increases at the same time, the response time of the recommendation system does not increase linearly with the increase in the number of recommendation requests, which means that even when the number of user visits surges, the recommendation system can provide recommendations smoothly. Serve.

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 belongs to the technical field of personalized recommendation, and particularly relates to a real-time hybrid recommendation method and a system based on a Labda architecture. The recommendation method comprises the following steps: providing recommendation service in a weighted hybrid recommendation mode by combining a content recommendation algorithm based on relative word frequency and a collaborative filtering recommendation algorithm based on incremental updating; meanwhile, updating the similarity between items in real time, adjusting candidate items in the recommendation list in real time, and enhancing the real-time performance and accuracy of a recommendation result; the system comprises an offline calculation module, a real-time calculation module and an external service module. The data source comprises news URL data and user click stream data; the user clicks the data to continuously reach the recommendation system in a stream form; according to the method, the advantages of the open source component are utilized, massive offline data are processed by means of a MapReduce programming thought, real-time incoming stream data are processed based on a Storm distributed stream computing framework, an offline computing part and a real-time computing part are combined, and finally the purposes of high throughput and low time delay of a real-time news recommendation system are achieved.

Description

technical field [0001] The invention belongs to the technical field of personalized recommendation, and in particular relates to a real-time mixed recommendation method and system based on Lambda architecture. Background technique [0002] In recent years, with the development of the Internet, the reading rate of traditional paper newspapers has been declining year by year, while digital reading has developed rapidly. Compared with traditional paper newspapers and periodicals, online news websites allow people to browse news and current affairs hotspots according to their own interests, and swim in the ocean of knowledge and information. But while online news sites have made information accessible, they have also created a certain amount of annoyance. The media constantly produces new news to meet people's need to pay close attention to current hot spots and follow-up reports of related events. As the number of news publications increases at a faster and faster rate, peopl...

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): G06F16/9536G06F16/9535G06Q30/02
CPCG06F16/9536G06F16/9535G06Q30/0255G06Q30/0271
Inventor 张鹏王海明顾宁卢暾
Owner FUDAN UNIV
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

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap