This is a never-ending discussion in our community. As a member of this community, I’ve tried to do a small contribution to this discussion by analzying co-authorship graphs of software engineering conferences. Results have been published in the paper: Analysis of co-authorship graphs of CORE-ranked software conferences.
More specifically, we define a suite of metrics for co-authorship graphs to study and characterize the community behind the software conferences included in the CORE ranking list. The graphs are built by mining the metadata provided by DBLP, an online reference for bibliographic information on major computer science publications. Metrics are first applied to each conference individually regardless their CORE rank and we then perform correlation analysis among the metrics. After, we study whether conferences ranked differently also present significant differences on their metrics values. This is done by analyzing variance complemented with factor and regression analysis.
Below you’ll find the abstract of the paper and the complete pdf.
Abstract
In most areas of Computer Science (CS), and in the software domain in particular, international conferences are as important as journals as a venue to disseminate research results. This has resulted in the creation of rankings to provide quality assessment of conferences (specially used for academic promotion purposes) like the well-known CORE ranking created by the Computing Research & Education Association of Australasia.
In this paper, we analyze 102 CORE-ranked conferences in the software area (covering all aspects of software engineering, programming languages, software architectures and the like) included in the DBLP dataset, an online reference for computers science bibliographic information. We define a suite of metrics focusing on the analysis of the co-authorship graph of the conferences, where authors are represented as nodes and co-authorship relationships as edges.
We aim to first characterize the patterns and structure of the community of researchers in software conferences. We then try to see if these values depend on the quality rank of the conference, justifying this way the existence of the different classifications in the CORE ranking system.
Full text
