Usage Count: A New Indicator to Detect Research Fronts

This article takes the regenerative medicine domain as an example to retrieve data. After using the topic search “regenerative medicine” in titles, abstracts or keywords (including keywords plus), and filtering out less representative record types, such as proceeding papers, meeting abstracts, news items, letters, etc., a total of 10,545 records dated between 2000 and 2015 were downloaded from WoS. We gathered the top 2,000 records sorted by both times cited and usage count. We regard these records as a representative dataset that can reflect the total records downloaded. The exported data format, search strategy, indexes, time span, and document types of the two indicators remained consistent.

Co-citation analysis is the typical bibliometric method, initially proposed by Small and Griffith (1974). Articles were clustered together based on their co-occurrence in the references lists of papers. In other words, if articles A and B are both cited by article C, it is more likely that they belong to the same research field and share similar topics or methods.

To identify clusters which represent the intellectual bases, spectral clustering algorithms have been used in this paper. Spectral clustering is a clustering method that uses eigenvectors of an affinity matrix derived from the data (Dhillon, 2004), and results derived by spectral clustering often outperform the traditional algorithms such as k-means or single linkage (von Luxburg, 2007). Besides, spectral clustering is easy to implement. This study uses CiteSpace 4.0.R5 (Chen, 2006) to identify the co-citation clusters and furthermore to find the citing articles of each cluster as well as the research front.

There are 84,315 and 113,339 references in dataset “times cited” and dataset “usage count,” respectively. In order to eliminate records that have little relationship with our research and pick out the most frequently used references, we select the top 1% most-cited records within the datasets to be further analyzed. In order to reduce disparities caused by absolute frequencies, we construct the co-citation matrix in terms of cosine coefficients. Additionally, a minimum spanning tree network is used in the software for network pruning in order to hide relatively weak citation links between item i and item j in the matrix, and improve the pruning efficiency as well. We compare the average publication year of citing articles (the recentness) as well as the mean cited year of the co-citation clusters. In this paper, average publication year of citing articles is equal to the mean publication year of the citing publications, and does not take the number of citations into account. Figure 1 presents organization of this study.

Figure 1

We investigate the distribution of citing articles and cited references within the two datasets. Figure 2 shows that the citing articles of usage count have experienced an exponential growth, because researchers prefer to use newly published literature, while Figure 3 indicates a right-skewed distribution for citing articles of times cited. This phenomenon is an indication that it will take several years for a paper to become highly cited. The distribution of cited references in Figures 4 and 5 presents a right-skewed distribution, but the data collected by usage count present a relatively high real-time property when compared with that collected by times cited. Many cited references were gathered from 2010 to 2013 by usage count in comparison with 2005–2008 by times cited.

Figure 2

Figure 3

Figure 4

Figure 5

There is approximately a four-year time lag in the mean publication year of citing articles, as well as a three-year time lag in the mean publication year of cited references collected by times cited. As is shown in Table 1, the recentness in times cited is 2009 and 2013.3 in usage count. As for the mean publication year of cited references, the figure is 2002.6 in times cited and 2005.7 in usage count.

Table 1

Overview of the dataset.

In this section, we compute the recentness of each research front. Table 2 lists the details of the two networks in which the number of articles in each cluster is above five, and the study ranks each cluster by recentness. The method to label the cluster is based on word profiles derived from the title of an article which cites the most co-cited articles within the cluster (Chen, 2006).

Tables 2 and 3 show there are 20 research fronts detected by the times cited, and 26 detected by usage count. The recentness in a majority of clusters detected by usage count tends to be published within the last four years, while this figure is relatively less recent in those detected by times cited. The recentness is 2011.59 detected by usage count and 2009.07 by times cited.

Table 2

Recentness of the research fronts detected by times cited.

Table 3

Recentness of the research fronts detected by usage count.

As indicated in Tables 2 and 3, a majority of research fronts generated by usage count tend to be more newly published compared to times cited. Because usage count is a reflection of researchers’ interest level within the last 180 days, most researchers pay more attention to achievements published within the previous two-three years, in order to stay in step with their colleagues and keep abreast of what is going on across scholarly communities. Besides, usage count can capture users’ full-text searching behaviors instantly instead of waiting for the tedious publishing process compared with times cited. Due to this, it is reasonable that a large proportion of citing articles in each cluster created by usage count are newly published. Meanwhile, we also observed that articles detected by usage count were published almost two years later than those detected based on times cited accordingly, because the newly published citing articles are prone to cite recent achievements due to rapid knowledge updates in the regenerative medicine field.

The two indicators both detect the induced pluripotent stem cell (IPSc) as one of the research fronts in the regenerative medicine field. We calculated all the citing articles of the IPSc field detected by the two indicators (Table 4 only lists the top 10 citing articles), to compare the recentness of the common research fronts detected by usage count and times cited. There are 55 citing articles in the IPSc field detected by times cited and 22 by usage count. The recentness in the IPSc field created by usage count is 2011.09 and 2010.07 by times cited. Moreover, the two indicators found seven common papers in the IPSc field.

Table 4

Comparison of the top 10 citing papers of the common research front.

As illustrated in Table 4, the two indicators can both detect research fronts in the IPSc field, and the citing articles of the IPSc field detected by usage count tend to be published more recently than times cited. Takahashi and Yamanaka (2006) originally introduced IPSc in 2006, and showed that the introduction of four specific gene encoding transcription factors could convert adult cells into pluripotent stem cells, and was awarded the 2012 Nobel Prize. As the shortage of donor organs for treating end-stage organ failure highlights the need for generating organs from IPSc (Takebe et al., 2013), we can expect that more and more researchers will retrieve and download classical articles in this field, and thus usage count will capture and accumulate the usage logs accordingly. Articles listed in the IPSc field by usage count are expected to be more recenly published than those listed by times cited.

In this section, we compare the top 10 most highly cited papers detected by usage count and times cited. Frequency refers to the times cited in local datasets. The results indicate that there are four papers in common listed in Table 5. Coincidentally, these papers rank within the top five results due to frequency. Moreover, there are three articles published before the year 2004 detected by times cited, while no article is published before 2004 in the top 10 detected by usage count.

Table 5

Top 10 highly cited papers detected by usage count and times cited.

From Table 5 we can see that the top 10 most highly cited papers selected b usage count and times cited are classical articles, but we find papers selected b usage count tend to be more recently published than those by times cited. The mea year of the top 10 most highly cited papers is 2004.6 detected by times cited, whil this figure is 2007.2 when sorted by usage count. It indicates an approximate three year time span among the mean cited years (known as “intellectual base”) of al clusters generated by usage count in the regenerative medicine domain.