Mechanical Categorization of Open Source Projects

Open source projects with open source software (OSS) have permeated society, and no industry can exist without OSS. At the same time, the risks and life cycle of OSS have severely impacted business and social life. Therefore, organizations utilizing OSS have started to analyze trends of OSS and use the analysis in their own organization's strategy. In an effort to look at OSS from a bird's eye view, OSS classification is important for human recognition. Without the OSS category, important OSS for the organization's business will be missed among too many OSS. At present, as multiple initiatives coexist, each uses its specialized OSS category, even in a single organization. For a long time, OSS stakeholders have manually organized OSS categories, and they have put in a great deal of effort. If manual work continues in the future, much effort will be required in the rapidly changing world of OSS. This research is fundamental and widely available for every analysis related to OSS, providing a universal methodology for OSS categorization with high validation accuracies of 0.98 by means of artificial intelligence technology. During this execution, there was an issue that the accuracy was low due to the small amount of suitable training data for OSS, but it was resolved with data augmentation.