The scientific activities carried out by the doctoral scientists from various facets of science were based on worthy knowledge which are technology driven and this had brought about many innovations and yet more of it to come. These discoveries and innovations do not just come to being by chances; rather utilize existing knowledge in new scientific investigations (Burnie, 2008). This existing knowledge are often documented and communicated in the journal articles. Consequently, the advent of internet has tremendously made publishing of digital content more ever accessible; making spontaneous influx of scientific communication from diverse group of interacting scientists who are geographically dispersed forming scientists' social relationships (Estabroks et al., 2008; Wang, Jiang & Ma, 2010).
As scientists engaged in active research during the course of their doctoral studies in the knowledge generating process, it is an herculean task for the researcher to dissect related published articles by other researchers in related fields (Su et al.,2009). Hence, "tracking researchers and articles in related fields are vital activities for the academic community" (Su et al., 2009, p.4287). Thus, there is clear need to explore the citations made in science doctoral students' theses in Faculty of Science, University of Ibadan to reveal the intellectual structure of science journals--depicting the connection and relationship within the disciplines.
The University of Ibadan (UI) started off as the University College, Ibadan (UCI) which was founded in 1948. It was the first University in Nigeria which was then a College of the University of London in a special relationship scheme, until it became fully independent in 1962. The Faculty of Science of UI happens to be one of the three foundation faculties at the inception of the University. The foundation departments under Faculty of Science: Archaeology (later combined with Anthropology), Botany and Microbiology, Chemistry, Mathematics, Physics and Zoology. In later time, other departments: Geology, Statistics and Computer Science were established in 1959, 1965 and 1974 respectively.
Citation analysis is a technique used to explore the underlying methods and behaviours of authors toward referencing the works of others, in the course of doing theirs (Eyong, 2010). Using citation analysis in research evaluation is often based on citation counts, presuming the publications on the reference list are of quality and have impact on the author that cited such publications, and perhaps have intellectual influence (Smith, 1981).
The citation network starts to build up when an article acknowledge the use of other earlier articles by referencing them (Havemann and Scharnhorst, 2012). This shows the existence of relationship among the referenced articles that influence the citing article. Thus, such cited works are said to be co-cited and the co-citation is based on the frequency of co-occurrence of the cited works in the same reference. When two or more articles shared one or more common references, such articles are said to be bibliographically coupled or co-referenced. The strength of connection in co-citation analysis is measured by the number of co-cited documents; the more co-cited documents are, the higher their co-citation strength. In contrast to bibliographic coupling which is measured by the number of shared references, the more shared references they have, the stronger their connection (Eom, 2009). In science mapping, the duo methods have been standard tools that tend to measure the degree of relationship or association between cited and co-referenced works for discovering the cognitive structure of research area and changes in patterns as the interests in the field changes (Eom, 2009; Sandor, 2014).
Co-citation analysis is a core element of citation analysis that examines the co-cited counts to reveal intellectual structure of many disciplines and keeps track of the evolution and impact of scientific knowledge over time (Zavaraqi, 2010). Though co-citation map has been rampantly used in exploring intellectual structure because of difficulties with extraction of bibliographic coupled counts from heavily data source such as Institute for Scientific Information (ISI) (Zhao and Strotmann, 2008b). Few works on citation network analysis have been based on bibliographic coupled counts (Zhao and Strotmann, 2008a), such as Jarneving (2007), Zhao and Strotmann (2008a) and Sandor (2014).
Citation network analysis can be performed at different units: author, journals, keywords etc., but it is mostly often done at author's unit than other units (Astrom, Danell, Larsen, & Schneider, 2009; Chen and Lien, 2011; Jarneving, 2007; Nerur, Rasheed, & Natarajan, 2008; Olatokun and Makinde, 2009; Pilkington, 2006; White & McCain, 1998; Zhao and Strotmann, 2008a). For this study, the unit of citation analysis is the journal.
Why journal as a unit of analysis? As revealed from the pilot study, the use of author as the unit of analysis for this study tends to widen the possibility of citing dissimilar authors in various science disciplines because of the authors' specialties that differs within its discipline and across the entire science disciplines. Instead, the use of journal as the unit of analysis encompasses and accommodates enough articles from various authors from diverse subject areas within the same specialty and related ones.
Numerous citation network analysis studies had been done in various fields such as information science (White and McCain, 1998), democracy-related (Liu and Wang, 2005), organic chemistry (Jarneving, 2007), knowledge utilization (Estabroks et. al, 2008), and operation management (Pilkington & Meredith, 2008). In Nigeria, myriad of citation analysis studies were carried out on students' academic research works in different fields: citation analysis on bachelor degrees' projects (Nikko and Adetoro, 2007; Iroaganachi, Iteskor, & Ifeakachuku, 2014, etc.); on masters' dissertations (Okiy, 2003; Aina, 2006; Ejekwu, 2010, etc.), and on doctoral theses (Olatokun & Makinde, 2009; Donatus (2010); Eyong, 2010, etc.). Only Olatokun & Makinde (2009) conducted author's cocitation map of cited journals in doctoral theses in animal science department. By and large, studies mapping the intellectual structure in other disciplines are still lacking.
This study tends to fill the literature gap with the aim of using citation data to map the intellectual structure of cited journals in science doctoral theses in University of Ibadan in directed network and to unveil the shared journal units among these science disciplines. This will give a broad understanding of the science doctoral theses' journal citation space--corresponding relation and connection between science disciplines. Precisely, it presents the visual and quantitative understanding of the interactions between cited journals in the science disciplines (Liu and Wang, 2005). This study is aimed to address the following research questions below:
i. What are the structural characteristics of journal citation network of science doctoral students?
ii. What are the subgroups that constitute the intellectual structure of science doctoral students' journal citations?
iii. Which of the journals are the most influential on the science disciplines?
iv. What are the percentages of shared or coupled journals among the science disciplines?
In citation networking analysis, articles are referred to as nodes or vertices and citations are the links or edges that connect the corresponding author. In view of this, the following citation network key terms of interest are described below.
Node: is a vertex that represents an individual such as journal, author, institution etc. within the network. Node is also called an actor.
Network size: is the number of nodes in the network.
Tie: is also referred to as an edge or line that links two or more nodes in the network together. In the network connection, if the edges indicate direction such network is called directed (asymmetric) network and if the edges are not directed, is called undirected (symmetric) network. The number of possible ties is k*(k-1) for directed network, and for undirected network is k*(k-1)/2, where k represent number of nodes present. The directed network indicates strong connection while undirected network indicates weak connection (Golbeck, 2013).
Degree: is the number of edges connected to a node. If a node has no connected edge, then the degree of the node is zero. Whatever measures of degree calculated in a directed network, it computes in twosome: total number of ties sent (outdegree) and ties received (indegree).
Density: is the ratio of available connections or edges in the network to the number of all possible connections within the network. For directed network, the density can be expressed as l / (k*(k-1)) where l is the number of the lines or edges present and k represent number of nodes present. The more connections within the network, the more densely it becomes (Jo, Jeung, Park, and Yoon, 2009).
Geodesic: is the shortest path between two or more nodes. The geodesic length between those nodes is referred to as geodesic distance. In strengths of ties measures, the distance between two nodes defines the strengths of the shortest path between them (Suerdem and Bicquelet, 2014).
Cohesion: measures how well the network is connected. It examines compactness of the network through testing the cut-point in the network, this is based upon distances between the nodes.
Centrality: it measures how many connections one node has to other nodes, i.e. how the node dominate over others in the network (Suerdem and Bicquelet, 2014). The centrality could be measured in terms of the degree, closeness and betweenness of the node in the network.
Degree centrality: is the degree of the node. The more the direct connection to the node, the better it is. Any indirect...