Research And Implementation Of Drug Recommendation System Based On Knowledge Graph

In recent years,the number of online pharmacies in China has been growing rapidly,and when users purchase drugs from online pharmacies,there is a wide variety of drugs,and users know little about the drugs and disease expertise.It is difficult to find the medicine you need quickly and accurately when you are shopping for it.That’s why a good drug recommendation system is crucial for online pharmacies.In the existing online pharmacies,some recommendation systems have been applying,such as direct sorting recommendation by drug sales,but such a recommendation method obviously cannot play a good role in assisting drug purchase.To address this problem,a drug recommendation algorithm integrating knowledge graph and collaborative filtering,is proposed in this thesis.And a personalized drug recommendation online pharmacy system matching with the algorithm is implemented.The main research work of this thesis are as follows.(1)Construction of the knowledge graph in the pharmaceutical field.Python’s Scarpy framework is used to crawl drug-related data information from three medical and health websites(doctor-seeking drug network,family doctor network,and 39 health network),and the drug knowledge graph construction structure of OMAHA(Open Medical and Health Alliance),a professional medical information service platform,is used to abstract drug domain knowledge graph entities and relationships in a top-down manner,followed by knowledge extraction and knowledge fusion to obtain drug knowledge graph triad information,and the triad information is stored in a neo4 j graph database to complete the construction of the drug domain knowledge graph.(2)a drug recommendation algorithm fusing knowledge graph and collaborative filtering.This thesis introduces the semantic information between drugs as an important recommendation basis by constructing a drug knowledge map,and the knowledge representation learning translation model Trans R maps the drug knowledge map into a low-dimensional continuous vector space,from which the semantic similarity between drugs and their relationships are calculated,and the fusion similarity is derived by combining the behavioral similarity derived from collaborative filtering,and the recommendation list is generated by predicting the score based on the fusion similarity.(3)Design and implementation of personalized drug recommendation online pharmacy system.Based on the above algorithm model,the B/S architecture is used to develop the personalized drug recommendation online pharmacy system in combination with the development requirements of the online pharmacy system,Vue3 is used for the front-end of the system,Spring Boot,Spring Cloud and other technologies are used for the back-end,and the overall system is built in the form of microservices by dividing the modules according to the business.The integration of drug recommendation algorithm and online pharmacy system is completed.