Open Source Software (OSS) are continually received more consideration from the software industries, since its value has been proved. There is much research aimed to understand the activities in the OSS project societies. ``Patch submission and acceptance" is known as one of the most important activities involving with all the OSS participants. Unfortunately, the existing patch-related studies may not be feasible due to an inappropriate simplification in the patch detection procedure. It would suffered the previous study results to be inaccurate similarly to how a well-recognized under-fitting problem degraded a prediction model, thus the conclusive comprehensions from the too-simplified patch study models such as a report that small patches were more preferable in the OSS community would be no longer applicable.
To overcome the problem, in this thesis, a fine-grained algorithm is introduced. The algorithm includes the partially and gradually accepted patches conditions that were overlooked but they regularly occur in actual. The results derived from the fine-grained algorithm would be more realistic giving better comprehension for all the OSS participants. To validate the algorithm, a patch analysis framework was constructed and deployed two large case studies using two well-known OSS project repositories named Apache Http Server and Eclipse Platform. The results demonstrated the case studies can lead the practitioners to the new comprehension of patch-related activities in OSS.