Research On Statistical Testing Methods Of Web Applications Based On Association Rules

With the increasing complexity of the web application structure and environment,in order to ensure that the web application software design meets the expected functional requirements,web statistical testing can effectively ensure the quality of Web applications within a certain time and investment.However,there are still some shortcomings in the traditional statistical testing.First,traditional statistical testing are not clear enough to quantify user behavior in web logs.How to extract and separate more valuable user information from web server logs is called an important part of web statistical testing.Secondly,the focus of the statistical testing is not prominent.There are many web application site pages.How to remove the pages with few visits,low relevance,and low interest.Especially How to mine frequent itemsets more effectively in the face of large data volume of web logs.In addition,the test cases generated by traditional statistical testing cannot simulate user behavior well.For the above problems,this paper proposes a test case generation method and reliability measurement method for web statistical testing based on association rules.For the quantification of user behavior in the web log is not clear enough,this article first extracts the corresponding fields from the Web server log,such as Referrer and URL,and saves them in a custom data structure note based on the hash,and generates a note list to improve query and comparison efficiency,and more detailed records of every user behavior.In order to remove web pages with few visits and low interest from the pages of the web site,and conduct more targeted testing of web applications,this paper proposes the Apriori algorithm based on the note_tree structure,then mines the note list to obtain frequent visits by users.In web statistical testing,Markov models are generally used for system modeling.The generated Markov models use roulette algorithms to generate test cases.This paper implements a test case generation algorithm for web statistics testing based on roulette algorithm,which takes into account both the size of the transition probability in the Markov model and the randomness of selection in order to be closer to real user behavior and further improve the test case.According to the generated test cases,the system reliability assessment was performed using the Nelson model,and MTBF was selected as the system reliability assessment index.Finally,this paper proposes three more comprehensive questions based on research and conducts related experiments.The experiment proves that the Apriori algorithm based on the note_tree structure has a shorter running time than the Apirori algorithm,and generates fewer frequent 2-itemsets L2.The Apriori algorithm based on the note_tree structure is more efficient in mining large amounts of web logs.The web application statistical test method based on association rules can more accurately measure the reliability of web applications.The MTBF value of the web application calculated by executing the test cases is similar to the MTBF value in the real environment,which verifies the effectiveness of the method.