Research And Application Of Intelligent Recommendation Methods For Internet Of Vehicles Service

The Internet of Vehicles is an important direction for the development of the automotive industry in recent years.It integrates vehicles into the ecosystem of the Internet of Everything,making vehicles a new communication terminal for users and services.The problem of service recommendation in the Internet of Vehicles scene has become a research hotspot.Many researchers have migrated the traditional Internetoriented intelligent service recommendation algorithm to the on-board intelligent terminal scenario,forming a new Internet of Vehicles service recommendation method.However,the Internet of Vehicles scenario brings new challenges to the above algorithms.Traditional Internet service recommendation platforms use user behavior(accessing pages and performing operations)to model user intent and preferences,while user behavior in the current vehicle environment is expressed through trajectory data.,Unable to obtain user information directly.Therefore,analyzing user behaviors based on trajectory data to obtain users’ interest points,preferences and residence time,thereby recommending personalized services to different users has important research and application value.This paper will start from the trajectory data collected by the terminal equipment of the Internet of Vehicles,and obtain the user’s behavior and intention through the extraction of user and location characteristics,and conduct related research on intelligent service recommendation.The research of this article will start from the following four aspects:(1)In order to study the user’s personalized characteristics,this paper designs an efficient parking spot recognition algorithm to show the user’s travel intention.The density clustering is used to study the user’s travel rules,combined with the open source interface to assign industry category attributes to the user’s points of interest,and the trajectory data is transformed into semantic user travel records.(2)Aiming at the complexity of user characteristics,this article summarizes the information that location data can display,defines a user portrait model,which is composed of three aspects: user basic information prediction,driving data statistics,and important location information identification.Design and implement calculations The methods related to user attributes are stored in the database.(3)In order to in-depth study the user’s travel intention,so as to provide real-time service recommendation to the user while driving,this paper conducts the relevant research on the user’s next location prediction.Based on user travel record data,a deep learning model is established to characterize the time,space,and address attribute information in the data.Through comparative experiments with multiple algorithms,the accuracy and usability of the model in this paper are verified.(4)Since users usually drive a car with a clear purpose,service recommendation should focus on the rationality of the location,so this paper designs a method to predict the user’s route area.Based on the environment,real-time scenes and user characteristics in the route area,a prediction vector is constructed to express the user’s intention,and address service recommendations are made.Monitor vehicle driving status and road condition information in the route area,and push information on traffic situation,vehicle safety,etc.In order to facilitate the analysis and management of various data information and verify the feasibility of the theoretical method,this paper designs and implements an intelligent recommendation system for the Internet of Vehicles service,and tests the system.