| Summary: | 碩士 === 中原大學 === 資訊工程研究所 === 105 === GitHub is the largest open source platform in the world. It is not only a famous social website about sharing coding, but also provides individuals or enterprises a stage to show their abilities. If the job seeker add the personal GitHub account into the resume, the recruiting enterprise can observe his performance and social network in the participation of collaborative coding. In the future this may form a new recruiting channel. However, for the human resources department or the interview officer, observing how the job seeker behaves on GitHub in order to make a judgement requires various domain knowledge. Moreover, it may spend a lot of time and manpower. In this study, we analyze the personal ability in expertise, divided into two categories, the degrees of contribution and the influence. They respectively correspond to his own effort and the assessment from others. We propose the approaches to extract the event log data of GitHub Archive and estimate the personal ability in expertise. After that, social networks for analysis are constructed. Furthermore, we base on the link analysis to design the methods for computing the degree of contribution and the influence. From the experiment for estimating the personal ability in expertise, comparing the list sorted by estimated values and the one sorted by real values, we find that the former indeed gives priority to better candidates. In the experiments of ability computation, according to the feedbacks from real users in GitHub, the sorted results in the ability of contribution can achieve 72% precision, and those in the ability of influence can even achieve 81% precision.
|
|---|
