A fast recommendation method in an online social network tagging system

Abstract

The invention discloses a fast recommendation method in an online social network tagging system, belonging to the fields of personalized recommendation and data mining. It solves the problem that traditional algorithms either one-sidedly pursue high accuracy while ignoring calculation cost, or one-sidedly pursue simple model and fast calculation while ignoring accuracy and personalization. The rough clustering algorithm of the present invention does not need to iterate to converge, which is essentially a process of quickly dividing users into clusters according to the user similarity index, and the user-based collaborative filtering recommendation algorithm is also recommended according to the similarity between users, so this The invention can ensure high accuracy and personalization; thereby reducing the calculation time overhead on the premise of ensuring accuracy and personalization; on the basis of ensuring the accuracy and personalization of the recommendation algorithm in the recommendation system, reducing the system time overhead, At the same time, it has the characteristics of high accuracy, high personalization and low system time overhead.

Description

technical field [0001] The invention belongs to the fields of personalized recommendation and data mining. Background technique [0002] Acronyms and key term definitions: [0003] Accuracy: Refers to the similarity between the resources or items recommended by the recommendation system and the resources or items actually liked or purchased by users in the system. This indicator is used to describe the recommendation ability and actual utility of the recommendation system. [0004] Personalization: Refers to the recommendation system recommending corresponding resources or items according to the different characteristics of each user. Personalization is highly related to accuracy. [0005] System time overhead: refers to the sum of the time overhead for the recommendation system to perform data cleaning based on the existing data in the online system and the time overhead for calculation based on the cleaned data and related recommendation algorithms. [0006] Timeliness: ...

AI Technical Summary

Problems solved by technology

[0011] Aiming at the deficiencies of the prior art, the present invention provides a fast recommendation method in an online social network tagging system, which solves the problem that traditional algorithms either pursue high accuracy and ignore calculation costs, or pursue simple models and quickly calculate while ignoring accuracy Rate and Personalized Questions

Images