Smart TV user behavior acquisition method and system based on exception mining algorithm

Abstract

The invention discloses a method and system for acquiring user behavior of a smart TV based on an exception mining algorithm. The method includes: when it is detected that the smart TV is powered on, collecting a user feature vector for determining user data when the user watches TV, and converting the user feature vector Regularization, and hash dimensionality reduction, to obtain dimensionality reduction user feature vectors; according to the hybrid algorithm of K-means clustering algorithm and hierarchical clustering algorithm, dimensionality reduction user feature vectors are divided to obtain multiple clusters corresponding to different K values Class tree, and obtain the clustering tree corresponding to the K value when the Gini impurity is the smallest as the optimal clustering tree; when the distance between the centroids of each cluster in the optimal clustering tree is greater than the preset distance threshold , then save the user feature vector corresponding to the optimal clustering tree. The invention realizes the identification of rare users, expands the diversity or coverage of content services, and at the same time has higher calculation efficiency and better clustering effect in the identification process.

Description

technical field [0001] The invention relates to the technical field of smart TVs, in particular to a method and system for acquiring smart TV user behaviors based on an exception mining algorithm. Background technique [0002] The purpose of smart TV user behavior analysis is to mine the behavior characteristics of smart TV users, understand the user's viewing habits, and provide users with valuable program content according to user needs. If the quality of the user behavior analysis is high, and the favorite TV programs and products are recommended to the user, then the user will become dependent on the smart TV. In order to strengthen content services and provide users with personalized services, it is necessary to understand user needs, understand user behaviors of using TV, and analyze user habits, so as to provide product planning and product positioning for the product planning department and provide users with better services. content services and personalized servic...

AI Technical Summary

Problems solved by technology

[0003] Most of the current user behavior analysis is based on clustering or classification algorithms to divide users into several types, and make corresponding product design or content services for different types of users. However, for some small number of users, the time and content of watching TV vary. For the majority of people (such as turning on the TV at 4:00 in the morning, watching mostly handball, ice hockey and other niche sports), it is difficult to be considered

Images