Design And Implementation Of Screening System For Android Performance Test Cases

According to the total annual mobile operating system market share report in 2020,Android system leads with 76.33%,followed by IOS with 23.49% share.Due to the increasing demand for android devices,major android manufacturers need to conduct a large number of tests on their performance in order to ensure the quality of their devices.The performance test cases are updated with the update of the application,and the long-term development will cause the maintainability of the performance test cases to become worse and worse.In the face of increasingly complex test cases,it is necessary to design and implement an android performance test case screening system to maintain the test cases,thereby improving the efficiency of performance testing and resource utilization.The main work of this thesis is as follows:(1)Introduce the subject background and research significance of this thesis;investigate and summarize the status quo of test case management platforms at home and abroad,analyze and summarize the defects in the existing test case management platforms;introduce related technologies used in the system;conduct demand analysis on the android performance test case screening system;complete the design of each functional module and database design according to the demand analysis.(2)The whole system is divided into four modules,and the core parts of the test case de-duplication and test case classification are designed with algorithms.The user management module realizes the management of user authority and user information.In the test case de-duplication module,aiming at the problem of insufficient test case context information,an incremental clustering algorithm T?Single-Pass that integrates test case features is proposed.First,perform part-of-speech judgment on the preprocessed test cases,select nouns as clustering objects,then select feature words based on the position of the words in the test cases,and finally use the manhattan formula combined with time factors to calculate the similarity between the test cases and the existing clusters to complete the incremental clustering of the test cases.Experiments show that the clustering algorithm used in this thesis has the best effect.In the test case classification module,in view of the problem that the unbalanced distribution of test case categories will cause errors in the multi-classification results,a multi-layer improved T?Fast Text classification algorithm is proposed on the basis of the Fast Text model.In the input layer,the N-Gram processed data is compared with the data in the Hanlp word segmentation dictionary,and the Bloom filter is used to filter;the hidden layer introduces the relationship between the in-classes,the inter-classes and the position;the output layer uses the case according to the feature words divide into multiple sentence blocks,and predict the category to which the entire test case belongs based on the category of each sentence block.Experiments show that the classification algorithm used in this thesis has the best accuracy and reliability.The test case search module uses Lucene technology to realize fuzzy search and precise search.(3)Combine the research and design of the above functional modules to realize an android performance test case screening system,and finally complete the functional test and performance test of the system.