No abstract

In this article we discuss the five yearly screenings for publications in questionable journals which have been carried out in the context of the performance-based research funding model in Flanders, Belgium. The Flemish funding model expanded from 2010 onwards, with a comprehensive bibliographic database for research output in the social sciences and humanities. Along with an overview of the procedures followed during the screenings for articles in questionable journals submitted for inclusion in this database, we present a bibliographic analysis of the publications identified. First, we show how the yearly number of publications in questionable journals has evolved over the period 2003–2016. Second, we present a disciplinary classification of the identified journals. In the third part of the results section, three authorship characteristics are discussed: multi-authorship, the seniority–or experience level–of authors in general and of the first author in particular, and the relation of the disciplinary scope of the journal (cognitive classification) with the departmental affiliation of the authors (organizational classification). Our results regarding yearly rates of publications in questionable journals indicate that awareness of the risks of questionable journals does not lead to a turn away from open access in general. The number of publications in open access journals rises every year, while the number of publications in questionable journals decreases from 2012 onwards. We find further that both early career and more senior researchers publish in questionable journals. We show that the average proportion of senior authors contributing to publications in questionable journals is somewhat higher than that for publications in open access journals. In addition, this paper yields insight into the extent to which publications in questionable journals pose a threat to the public and political legitimacy of a performance-based research funding system of a western European region. We include concrete suggestions for those tasked with maintaining bibliographic databases and screening for publications in questionable journals.

We compare two supervised machine learning algorithms—Multinomial Naïve Bayes and Gradient Boosting—to classify social science articles using textual data. The high level of granularity of the classification scheme used and the possibility that multiple categories are assigned to a document make this task challenging. To collect the training data, we query three discipline specific thesauri to retrieve articles corresponding to specialties in the classification. The resulting data set consists of 113,909 records and covers 245 specialties, aggregated into 31 subdisciplines from three disciplines. Experts were consulted to validate the thesauri-based classification. The resulting multilabel data set is used to train the machine learning algorithms in different configurations. We deploy a multilabel classifier chaining model, allowing for an arbitrary number of categories to be assigned to each document. The best results are obtained with Gradient Boosting. The approach does not rely on citation data. It can be applied in settings where such information is not available. We conclude that fine-grained text-based classification of social sciences publications at a subdisciplinary level is a hard task, for humans and machines alike. A combination of human expertise and machine learning is suggested as a way forward to improve the classification of social sciences documents.

This article presents a cohort analysis to study changes in the publication patterns of scholars working at a social sciences and humanities (SSH) university department or research unit in Flanders, Belgium. Starting from a comprehensive bibliographic database, we analyze the peer review status, publication language, publication type (journal article, book publication, or proceedings), and coverage in Web of Science (WoS) for publications produced between 2000 and 2014. Through a cohort analysis of the authors, a distinction can be made between effects that reflect changes in the characteristics of how researchers of comparable seniority publish (intracohort change) and effects that are due to the disappearance of researchers and/or introduction of new researchers (cohort succession). Our findings indicate that there is a trend across all five cohorts and in both the social sciences and humanities toward peer review, use of English, and publishing in WoS-indexed journals. While we witness clear intracohort changes, cohort succession effects are shown to be much weaker. The oldest cohort appears to maintain a traditional SSH profile, with lower shares of peer-reviewed publications, publications in English, journal articles, and publications indexed in WoS. As for publication types, all cohorts exhibit a slightly declining share of journal articles over time in favor of book publications, particularly in the humanities. The study shows that cohort analysis is a useful instrument to gain better insight into the evolution of publication patterns.

The predatory nature of a journal is in constant debate because it depends on multiple factors, which keep evolving. The classification of a journal as being predatory, or not, is no longer exclusively associated with its open access status, by inclusion or exclusion on perceived reputable academic indexes and/or on whitelists or blacklists. Inclusion in the latter may itself be determined by a host of criteria, may be riddled with type I errors (e.g., erroneous inclusion of a truly predatory journal in a whitelist) and/or type II errors (e.g., erroneous exclusion of a truly valid scholarly journal in a whitelist). While extreme cases of predatory publishing behavior may be clear cut, with true predatory journals displaying ample predatory properties, journals in non-binary grey zones of predatory criteria are difficult to classify. They may have some legitimate properties, but also some illegitimate ones. In such cases, it might be too extreme to refer to such entities as “predatory”. Simply referring to them as “potentially predatory” or “borderline predatory” also does little justice to discern a predatory entity from an unscholarly, low-quality, unprofessional, or exploitative one. Faced with the limitations caused by this gradient of predatory dimensionality, this paper introduces a novel credit-like rating system, based in part on well-known financial credit ratings companies used to assess investment risk and creditworthiness, to assess journal or publisher quality. Cognizant of the weaknesses and criticisms of these rating systems, we suggest their use as a new way to view the scholarly nature of a journal or publisher. When used as a tool to supplement, replace, or reinforce current sets of criteria used for whitelists and blacklists, this system may provide a fresh perspective to gain a better understanding of predatory publishing behavior. Our tool does not propose to offer a definitive solution to this problem.