Code bad smell detection method based on BP neural network

Abstract

The invention relates to a code bad smell detection method based on a BP neural network, and belongs to the technical field of computer software. The method comprises: firstly, extracting a code bad smell instance and label information; then, calculating whether a code bad taste measurement characteristic exists or not; combining the measurement characteristics with the extracted label information; constructing a training set in this way, the method comprises the following steps: establishing a neural network by adopting Keras, taking a training set as input of a neural network model to finishcode bad smell prediction output training, and finally taking a code bad smell instance obtained from a tested program as a code bad smell test set to be input into the trained neural network model,so that the model output code bad smell belongs to a certain type. According to the method, a code bad smell detection technology based on metric characteristics and a neural network algorithm are combined to detect different types of code bad smells in the data set, so that the detection accuracy and the F1 value are improved.

Description

technical field [0001] The invention belongs to the technical field of computer software, especially in the technical field of bad smell detection of codes. Neural Networks for Code Smell Detection. Background technique [0002] Good software design quality can make maintenance and reuse easier and more convenient, and if there are various bad smells in the code, it will inevitably lead to a decline in the overall design quality of the software. Code bad smell detection has always been a field of software engineering. One of the research hotspots, in which researchers have proposed some code smell detection methods based on various machine learning algorithms. Kreimer proposed an adaptive detection method, combined with the decision tree algorithm to detect two bad code smells of overly large classes and overly long methods. Khomh et al. build Bayesian belief networks from the definition of antipatterns based on a target problem metric and verify the God class in two open ...

AI Technical Summary

Problems solved by technology

The present invention aims at the single-type problem in the data set of code bad smell detection, which does not conform to the distribution of bad code smell in the data set in the actual software development process, and proposes two codes based on BP neural network for long method and feature attachment Dataset detection method after bad smell types are merged

Images