Are author, affiliation, and citation networks predictive of a journal getting blacklisted?
Open access journals are becoming increasingly viable publication venues for scientists, educational organizations, and government funders. Unfortunately, unscrupulous publishers have taken advantage of this trend by creating journals that are open access but predatory or of low quality [1]. Some services attempt to remedy this situation by providing a white list and blacklist of journals, manually vetted by experts. Two examples of these expertly curated lists are the Directory of Open Access Journals (DOAJ) and the Cabbell's journal blacklist and whitelist. However, how these organizations choose journals is poorly understood. It would be beneficial to understand these decisions and also it would be important to improve on the detection accuracy of these services. In this preliminary work, we codify the rules that the DOAJ purports to use for journal auditing and examine their effectiveness in telling apart blacklisted vs whitelisted journals [2]. We compare these rules to features derived from the author, organization, and citation networks. We show that by using a combination of the DOAJ rules and network features, we can achieve significantly higher accuracy in our predictions. Finally, we examine the features that are most predictive and discuss our next steps.