Optimization of regular expressions using competitive coevolutionary algorithm based on symbolic regression
Liliya Anatolievna Demidova, Nikita Andreevich Moroshkin · ITM Web of Conferences · 2025
The paper presents an algorithm for optimizing the structure of regular expressions of the Python programming language dialect of the re module. The optimization algorithm is implemented as a competitive coevolution algorithm based on the symbolic regression algorithm (the Gene Expression Programming algorithm will be used as an implementation of the symbolic regression algorithm). The paper proposes a pseudocode for the regular expression optimization algorithm as an abstract “black box” model, provides hyperparameters of competitive coevolution, as well as a function for assessing the suitability and reliability of individual algorithms within the coevolution. A comparative analysis of the results of running the GEP algorithms as part of coevolution and separately demonstrates the effectiveness of using coevolution as a method for optimizing regular expressions.