Copyright Protection of Computer Software Deep Learning based Patent Text Clustering

The calculation method of the program is the core part of the software function, while the copyright law cannot effectively protect the software operation method, but can only protect the software development. Therefore, the instrumental value of software, such as its operation method, creation process and concept expression, can be protected by patent, so as to improve the protection of software property rights. This paper focuses on the analysis of the legal protection countermeasures of computer software intellectual property, in order to further promote the sustainable development of the computer software development industry. Establish data intellectual property protection system through computer neural network. The protection of data property rights and intellectual property rights based on deep learning and computer aided design is established. Two methods are used to initialize the word embedding layer, one is to use the pre trained word vector, the other is to initialize randomly. Using the advantages of recursive CNN (convolutional neural network) in processing sequence data, it can automatically capture the order and other correlations between text sentences, and train patent text according to the results of topic clustering. The results show that the final convergence accuracy of the recursive CNN combined model is higher than that of the comparison model, and the accuracy of the recursive CNN combined model reaches 95.39%, which indicates that the performance of the combined model based on the computer-aided model is better than that of the single model. The establishment of data intellectual property protection system through computer neural network is more stable and effective. Therefore, this model improves the level of intellectual property protection of data property rights. © 2023 CAD Solutions, LLC.