Hierarchical feature similarity integration method for data sets based on deep learning
YU Xiao-li · Journal of Physics Conference Series · 2021
Abstract Under the background of the rapid rise of open-source software and the gradual popularization of various software development tools, a large amount of development activity data has been accumulated on the Internet. In the process of using these data to construct data sets, due to their poor traceability and narrow application scope, the quality of data in development activities is not high and the accuracy of analysis results is not high. The application of the hierarchical feature similarity integration method of data sets can make the multi-version and multi-level development smoother and more orderly. In this paper, a hierarchical feature similarity integration method based on hierarchical deep learning is proposed for data sets. Firstly, the dynamic mesh partitioning method is used to divide the sparse and dense regions in the space, which reduces the scale of data detection and shortens the execution time of detection. Then, through the hierarchical deep learning process, the professional knowledge and the distribution information of data attribute value are fused to realize the detection of discrete data in the database. Experimental results show that this method can accurately complete the detection of discrete data in the database in a relatively short time, and has more application advantages than traditional methods.