Key references
Article information
The reference section of an academic journal paper plays a crucial role in many quantitative analyses of journals including their evaluation. Databases such as Web of Science and Scopus and evaluation metrics such as the impact factor and the h index are all based on the references. Many innovative activities initiated by Crossref are also based on connecting a great number of articles through their DOIs (digital object identifiers) and references. Currently, references of articles provide a unique route for imbedding them in a network of a huge number of academic articles. By analyzing this network, one can obtain the essential information about journals and researchers.
Because the reference section has such a great importance, it is highly desirable for authors to put much effort into selecting appropriate references for their papers. In the old days, when scholars who wrote one or two papers a year got enough recognition, I believe most authors spent a sufficient amount of time writing their papers and selected their references carefully. In these days, however, scholars need to write a lot more papers on the average to get recognized as productive researchers and acquire external research funding, especially in the areas of science and engineering. Consequently, they tend to make less effort in writing individual papers than before. As far as the references are concerned, I think it often happens that many of them are copied from earlier papers on the same topic, or are chosen simply because they were written by famous people or were published in prestigious journals, despite of having little direct relevance to the citing paper. Moreover, it is not unusual to find misbehaviors or unethical behaviors in referencing, such as self-citing the authors’ own papers excessively and some number of people or journals teeming up to cite each other [1]. All of these are impairing the real meaning and importance of the reference section.
From this perspective, I would like to make a new proposal. Nowadays, many journals require authors to select a small number of keywords for their paper. What about also asking them to select a small number of key references pertinent to their paper? Every author has a few references which played an especially important role in performing their actual research. They might have hit upon their crucial ideas or have learned essential techniques and methods by reading them. It is important to select those references as key references, not just those written by famous researchers in the field or published in top journals. From this criterion, it is natural to forbid author self-citations. It is also recommended that review papers, which are usually cited more often than regular papers, are not included in the key references.
The maximum number of the key references, N, needs to be small to make their selection to be meaningful. In order to use the database of key references in constructing independent journal and author metrics, all journals participating in this proposal have to adopt the same or similar value of N. I feel N= 5 can be a suitable choice. As these data are accumulated, I think it will provide useful information on the practical importance of the references to actual research, which is independent from that obtained by considering the whole set of references. It may provide a new measure for evaluating journals and researchers from a different perspective. It also has an advantage that there is no variation among different academic disciplines, if the same N is chosen for all journals.
Journal metrics based on references such as the impact factor have contributed to the development of academic journals greatly. More recently, however, it has been pointed out that these metrics have some deficiencies and need to be used with much caution. New metrics based on the proposal made here can be a useful supplement to the existing metrics. I have to acknowledge, however, that it is always possible to manipulate any metric, such as by two people citing each other’s paper. Nevertheless, with the rapid development of information technology, it will be possible to detect many anomalous behavior patterns and exclude the corresponding articles from the database to make the metrics more reliable.
Notes
No potential conflict of interest relevant to this article was reported.