Research On The Recommendation Method Of Student Practical Training Direction Based On Knowledge Graph

In the era of explosive data growth,recommendation algorithms have gradually become popular in the big data environment.As a widely used information filtering method,the personalized recommendation system actively recommends content of interest to users.Effectively reduce the cost of decision-making time.At the same time,personalized recommendations can make recommendations more in line with user preferences,so that recommendations can achieve the effect of " different people have different portraits".The knowledge graph is an important branch of artificial intelligence in the data-driven era,and it is also the cornerstone of the cognitive ability of machines.Recommendation algorithms based on knowledge graphs can provide rich prior knowledge for recommendation algorithms to improve the accuracy,diversity,and interpretability of recommendations.Refer to the training method of undergraduate students in the School of Software of the North University of China.Each junior students is required to choose a practical training direction.However,every year,most students are confused when choosing the direction of practical training.This article is based on the actual situation.Build a knowledge graph based on the students information.Improve existing recommendation algorithms.A new personalized recommendation method is proposed and applied to the recommendation of students’ practical training directions.It avoids the problem of everyone blindly choosing the direction of the crowd and also avoids the situation that individual directions are unpopular or full,which has practical significance and application value.This paper focuses on the construction of student information knowledge graph and the improvement of personalized recommendation algorithms.The main work of this paper includes the following aspects :（1）Construct a knowledge graph of student information.After designing and sorting out the student interest questionnaire,data processing of more than 2,000 students in the three grades.Using the Neo4 j graph database to build a knowledge graph of student-related information.Based on the practical training direction set by the college over the years and combined with the latest development trend of the software industry to set the training directions.（2）Improve the recommendation algorithm.Collaborative filter recommendation algorithm is currently the most commonly used recommendation method.On this basis,firstly,uses Canopy and Bi-kmeans for mixed the data clustering.Considered that user interests will change.The TF-IDF method will increase the time coefficient to calculate the user interest preferences to get the fusion of user interest and mixed clustering model when calculating user interest preferences.And integrate the user’s personal attributes into personalized recommendations to obtain a new similarity calculation model.Finally,the constructed knowledge graph and recommendation algorithm are applied to the recommendation system for students’ practical training directions.（3）Design and implement the recommendation system for students’ practical training directions.The system displays the relevant information of the training and generates the user portrait.Takes the performance as the main recommendation basis,and the information of the statistical survey such as personal interest is used as the auxiliary recommendation.Generates a recommended list of training directions for the student users.