Design And Implementation Of Agricultural Information Personalized Push System

The personalized push of agricultural information is a new technology that reduces the information overload by regularly transmitting the information needed by users on the Internet.It is an important part of agricultural information construction,and it is also the country’s all-round implementation of “precise poverty alleviation”.An important tool for precision poverty alleviation.Under the influence of the agricultural information personalized push system,the agriculture-related users can greatly reduce the time and effort spent searching and querying information,and solve the problem of users quickly and accurately obtaining information in the information ocean.And relevant departments can promptly push information on agricultural policies,pest control,scientific planting and other information to farmers in need,and solve the problems of obstacles and lags in information transmission between government and farmers.In view of the particularity of agriculture and the better service to users,this paper has carried out research work in the following aspects:(1)Research on classification methods of agricultural information.The complexity of agricultural information is high,the attributes between crops are crossed,and the number of classification labels is not fixed.Therefore,this paper establishes an agricultural information multi-label classification model for agricultural information,dynamically updates iteration based on user feedback information,optimizes the model training,continuously enriches the label library,and improves the accuracy of multilabel classification.(2)User interest model research.In order to fully understand the user’s hobbies,the system analyzes the user’s registration information,mines the user’s interests and interests,and deposits the user’s long-term interest database to establish the initial user interest model.According to the user’s browsing,scoring and other feedback behaviors,the user’s interests and hobbies are mined through the analysis log,and the user’s shortterm interest database is stored,and a timer is added to periodically update the short-term interest database.The user’s long-term and short-term interest weights are combined,and the user interest model can be adaptively updated according to changes in the short-term interest base.Improve the personalization of the system,so that the pushed content is more in line with the user’s interest.(3)Develop an agricultural personalized push system.Combined with the Slope one collaborative filtering recommendation algorithm,you can accurately push the user’s favorite content.The agricultural information personalized push system uses mobile APP as the main display platform of the system,which makes the user easy to operate and quick to get started,and can view agricultural information anytime and anywhere.From the beginning of the concept,the theoretical value and practical value of the personalized recommendation system are analyzed.Then consider the realization of the system with the idea of "object-oriented",and divide the system into the above three parts to study.In the process of sub-module research,we learned simple deep learning and realized the training of classification model;built an adaptive user interest model;analyzed the advantages and disadvantages of the recommendation algorithm,and selected the appropriate algorithm as the basic algorithm of the system.Finally,in the realization process of the system,the complete system architecture is designed,and the construction of the cloud server is studied,and the mobile APP is developed.The implementation of the system completes the precise push function of the mobile APP,and the push effect is ideal.