Development Of Agricultural Information Personalized Recommendation System Based On Persona

As a national strategy,"Internet +" has had a profound impact in all walks of life.The combination of the Internet and agriculture has improved the efficiency of agricultural production and produced a new economic model.At the same time,a large amount of agricultural information has been generated on the Internet,but the lack of professional and in-depth research on this information has led to the occurrence of information asymmetry,this phenomenon reduces the value of information,so that users with real needs do not have access to practical data.Aiming at the low accuracy of the individualized recommendation of agricultural information in view of the insufficient research on user portraits and personalized recommendations in the field of agriculture,this paper focuses on the establishment of persona for agricultural users.According to the established agricultural persona,it recommends the required agricultural information to users and improves persona through user feedback.Strive for accuracy and real-time of recommended information.The main contents of this paper are as follows:(1)A regional group-specific persona update model is proposed.Given the regional characteristics of agricultural production and agricultural information,in the case of missing new user behavior data,first establish a domain-based group persona as the initial persona of the new user.,then updates persona by collecting user feedback information,and finally obtains specific persona.Experiments prove that the model solves the cold start problem in the recommendation system well.(2)An interest update model for agricultural persona has been established.By establishing the individualized recommendation model of agricultural information,the information is recommended to the user,and then collect the feedback information of the user as the basis for the persona renewal.When updating the agricultural persona,the user interest labels are divided into long-term interest labels and short-term interest labels,the short-term interest labels are updated by using a sliding time window algorithm,and the long-term interest labels are updated by using a forgetting function algorithm.Experiments prove that the accuracy rate of the long and short term conbined update model is higher than 80%,which is better than the single update model..(3)The personalized recommendation system of agricultural information based on agricultural persona is realized.The system mainly realizes the display of recommended information and the collection of user feedback information.The visualization part of the foreground is completed by Mini Program,and the system uses Spring Boot framework in the background.Due to the cross-platform nature of Mini Program,the system has good popularization and practicability.