Research On Personalized Recommendation System Of Agricultural Information Based On User Portrait

Under the Internet plus Internet plus Internet,the government has pointed out the direction of agricultural development,that is,Internet plus agriculture,and has played the dominant role of "Internet +",expanding public services to remote areas,putting the requirements of "information into households" into practice,laying the foundation for "Internet plus agriculture".Providing accurate online resources for farmers will become one of the mainstream services.This study introduces personalized recommendation system based on recommendation theory such as user portrait when providing online education or agricultural services for basic agricultural users,so as to improve the "information explosion" and "resource overload" of agricultural users in resource acquisition,improve the utilization rate of agricultural information resources,and achieve better agricultural service effect.Based on the traditional recommendation system based on user portrait,this paper introduces the time variable to build the personalized recommendation model of agricultural information resources.So as to improve the accuracy of personalized recommendation of agricultural information resources,make up for the lack of personalized recommendation.In this project,first of all,the acquisition of agricultural information resources and data mining,learn from the excellent experience at home and abroad,combined with the actual situation of agricultural information in China,build a relatively complete,professional classification of information resources.Accurate positioning of regional characteristics of agricultural production,the establishment of regional based group user portrait.As the recommendation system new user personality data base.Secondly,the recommendation information is extracted and classified,and the recommendation model of agricultural information resources is formed by using agricultural user portraits combined with context based time variables.In the process of operation,through a large number of data collection and algorithm analysis,the system accurately grasps user preferences,recommends agricultural information to users according to their needs,and meets the needs of users for agricultural information resources.Recommendation and recommender directly form a two-way interaction mode.According to the user's choice of recommendation information,the recommendation service is continuously optimized and the recommendation information is updated,so that the whole system is in a virtuous circle of optimization development.Finally,wechat software with wide coverage is selected as the presentation platform,and wechat applet with clear page and convenient operation is used as the entrance of the system to carry the operation of the program.The main research contents are as follows(1)Establish and optimize group user profile based on region.The new user's initial interest tag is set up by combining the regional characteristic crops,the temperature zone characteristic crops and the user's self selected interest content.As the new user personality data base of the recommendation system.To a certain extent,the problem of cold start of recommender system is solved.(2)The updating model of agricultural user profile interest label was optimized.Users' interests are influenced by time policy,geography and other factors.In order to reflect users' interests more accurately,the system divides user tags into short-term tags,long-term tags and periodic tags.Then,the sliding time window algorithm,forgetting function mechanism and time function are used to update the three tags.(3)A new time function is introduced into the design.In this paper,we design a time function to generate different periods for different regions,and combine the time function with the recommendation algorithm based on user portrait,so as to better analyze the user's intention and improve the accuracy of recommendation.(4)The agricultural information recommendation system based on wechat applet is developed,and the accurate recommendation function of the recommendation system is realized.