基于增强命名实体识别的开源许可证条款识别方法OA

Open-source licenses have developed significantly alongside the widespread adoption of open-source software,defining the rights and obligations of open-source software users.Developers who use open-source code without adhering to the corresponding license terms may face legal risks such as infringement.Therefore,identifying open-source license terms and understanding license content plays a crucial role in risk management and intellectual property protection.Existing license term identification methods suffer from limited applicability and insufficient accuracy.We propose a license term identification method based on enhanced named entity rec-ognition(LTNER)to achieve automated and precise identification of license terms in open-source licenses.The method constructs an open-source license term identification model,leveraging BERT's contextual capture capabilities and deep semantic understanding to flexibly analyze various license texts,and uses named entity recognition to determine the terms contained in the license.Experimental results show that the LTNER model outperforms the existing state-of-the-art tool LiDetector in terms of recall and F1 score,with a 14.68%increase in recall and a 6.71%increase in F1 score,validating the model's effectiveness in identifying open-source license terms.