Design And Implementation Of Tumor Knowledge Recommendation System Based On Text Classification

At present,with the increasingly severe form of tumor prevention and treatment in China,the demand of tumor patients for tumor knowledge is more and more urgent,but under the impact of "information overload",the traditional information retrieval methods have been unable to meet the needs of users.Under this background,the research on tumor knowledge recommendation system is imperative.Tumor knowledge recommendation system can help patients dig out the parts they need from the mass of medical information and indirectly improve the quality of life of patients.At the same time,the tumor knowledge recommendation system can also create a medium for the communication between doctors and patients,and help doctors understand the patient's condition through questionnaire and follow-up,so as to promote a good doctor-patient relationship.The problem of cold start is an important challenge in the development of knowledge recommendation system,and it is particularly apparent in the early stages of system operation.To address this problem,this paper proposes a cold start problem solving strategy based on LSTM text classification.Combined with LSTM text multiple classification algorithm,it can quickly predict patients' knowledge preferences and recommend interesting content to them based on their medical records and browsing behavior.First,this paper designs and implements a preliminary solution strategy for cold start problems based on text classification.By training the LSTM model,the disease classification prediction of tumor knowledge is realized,and combined with the ICD-10 code of the patient's medical record,a rough recommendation is completed.This strategy solves the cold start problem of the system.Then,in order to solve the problem of insufficient accuracy in the preliminary solution strategy,this paper further integrates the collaborative filtering algorithm,and proposes an optimized cold start problem solving strategy.By analyzing the similarity of the user's medical record and browsing behavior,the user's score for articles is predicted,And the score prediction results and LSTM model prediction results are fused to form the final mixed recommendation results.This strategy not only solves the cold start problem recommended by cancer knowledge,but also achieves a high recommendation accuracy.Based on the above strategy,this paper designs and implements a tumor knowledge recommendation system.The system can collect the user's browsing time,article score and other information and upload it to the server.It also stores the patient's medical record information.Based on the above information,the algorithm server can generate an article recommendation list,and regularly provide article push services to users through the WeChat public account.