A framework for integrating bibliographical data of computer science publications

In this paper, we propose a framework to integrate bibliographical data of computer science publications from heterogeneous digital libraries. The framework consists of three key components: publication collector, bibliographical parser and duplicated checker. In order to analyze efficiency of our framework in integrating data from heterogeneous sources, we conduct experiment with three different digital libraries: Microsoft Academic Search, CiteSeerX and DBLP. At this time, our integrated dataset contains 5.320.539 publications and 1.723.148 authors and their metadata. Our dataset increases quantity of rows and columns compared with the others. Thus, it could be published for other studies related to bibliographical data such as searching literature, ranking publications, identifying the research trend, mining the linking of articles.