Design And Implementation Of An Auxiliary Scoring Tool For Big Data Management Training

With the rapid development of the Internet,the education industry is combined with the Internet technology,and the online training platform of auxiliary course teaching emerge as the times require.The big data management training platform provides students with virtual experimental environment,automatic deployment of experimental environment,experimental resource monitoring and other functions to assist students in online training.However,how to let students get feedback on the correctness of the experimental code and other evaluation dimensions in the course training process,and how to quickly return the evaluation results for the practical training problems under the large amount of data has become a problem to be solved in this platform.An auxiliary scoring tool for big data management training which is based on the above background and requirements is designed and implemented.This auxiliary scoring tool provides evaluation solutions for multiple databases under different question types.At the same time,the auxiliary scoring tool analyzes and compares the key words,operation types,execution plans and other aspects of the execution of code to check the correctness of experimental methods,and combines the correctness of experimental results and the performance of experimental execution to give the score comprehensively.Compared with the traditional scoring methods such as experimental results comparison or code similarity comparison,the multi-gradient training scores given by this method are more reasonable and objective.In addition,the auxiliary scoring tool provides two fast comparison methods to solve the problems of large data amount and inconsistent record order in the result set of large data amount.One is fast comparison of result set based on multi-thread Simhash;The second is the proposed pre-located dynamical Bloom Filter algorithm,which further improves the search speed and reduces the misjudgment rate through the pre-located hash and the longest matching method,and realizes the sampling comparison of the result set by this algorithm.In order to cope with the pressure of the concurrent judgement of question requests,each functional module is deployed through the microservice architecture to expand the nodes according to the access pressure,so as to provide fast question response.Finally,we verify that the multi-threaded Simhash algorithm can effectively improve the speed of result set evaluation and comparison by comparing with Shingle algorithm and Simhash algorithm.And we verify that the pre-located dynamical Bloom Filter can effectively improve the query speed,reduce the rate of misclassification and optimize the space usage by comparing with other Bloom Filter algorithms.The pre-located dynamical Bloom Filter can effectively improve the speed of result set comparison.Through the design of Neo4 j and Mongo DB typical experiments to verify the integrity of each interface function available,the actual performance of the auxiliary scoring tool in line with the design expectations,and can form a multi-gradient score for different experimental results,can provide good assistance to the course training.