Personalized Recommendation Of Automobile For New User Demand

With the rapid iteration and development of Internet technology,the amount of information generated by human society in the past 30 years has greatly exceeded the total amount of information generated by human history.People are trapped in the ocean of information,how to get the information you need quickly becomes more and more difficult.By analyzing user behavior,personalized recommendation establishes interest model for users,predicts user behavior pattern,and then actively recommends effective information to users,thus to a certain extent,it solves the difficulty for users to obtain useful information.However,unlike the common commodities in people's daily life,automobiles,which are purchased at low frequencies,have the characteristics of high complexity and many related parameters.Therefore,in the process of recommendation task,users who do not often contact this field will encounter the lack of knowledge in related fields,and sellers of automobiles will encounter difficulties such as fewer historical transaction data of users.How to successfully apply personalized recommendation in the field of automobile purchase,make it easier for users to choose and buy automobiles,and make automobile sellers recommend automobiles more accurately is an urgent problem for users and automobile sellers to solve,and is also the purpose of this study.On the basis of relevant theories and research achievements of domestic and foreign scholars,this paper establishes automobile recommendation algorithm to meet the needs of new users.The main achievements of this paper are as follows:1.By comparing the existing recommendation algorithms,the framework of personalized recommendation algorithm,knowledge-based recommendation algorithm,is determined,and the algorithm and related technologies are analyzed and introduced in detail.2.Obtain the vehicle data needed by this paper through the web crawler,analyze the preferences of different groups for vehicle type,price and travel purpose,analyze the online comment text,and draw word clouds of different users and different vehicle types.3.Aiming at the new user's car-buying demand,we get the new user's car-buying demand through conversational interaction,and then get the new user's characteristics.Then we build the car recommendation algorithm based on Pearson similarity and cosine similarityrespectively,combining with the user data of the purchased car,which overcomes the problems of cold-start and sparsity in the new user recommendation.Based on the above research results,for new users,the application of this method makes it possible for users to quickly obtain useful information from the ocean of information.For car dealers,the application of this method greatly reduces the time and energy spent in recommending cars to users.Therefore,the research results of this paper have important theoretical significance and practical application value in the field of automobile purchase.