This document discusses mapping and visualizing the core of scientific domains using social network analysis techniques. It introduces the concept of a "Network of the Core" (NC) to represent relationships between theoretical constructs, models, and concepts. NCs can be directional, showing causal relationships, or directionless, showing general connections. NCs can reveal hidden characteristics of a research domain like central constructs. The document demonstrates directional and directionless NCs for information systems research domains. NCs help conceptualize domains, identify missing links, and explore research opportunities. Future work should construct more detailed NCs to analyze research domain structures.
1. Mapping and Visualizing The Core of Scientific Domains: Information System Research Authors: GoharFeroz Khan* Junhoon Moon** Han Woo Park* *Department of Media & Communication, YeungNam University, Republic of Korea **Information Management & Marketing, College of Agriculture and Life Sciences, Seoul National University, Republic of Korea Prepared for COLLNET 2011, Seventh International Conference on Webometrics, Informetrics and Scientometrics (WIS), 20-23 September, 2011, Istanbul Bilgi University, Istanbul, Turkey, http://collnet.cs.bilgi.edu.tr/program/programme/ An updated version of this article is accepted for publication in the Scientometrics journal
2.
3. Visualizing and gauging a network of scientific knowledge is an emerging area of interest (Blatt, 2009; Perianes-Rodríguez, Olmeda-Gómez, & Moya-Anegón, 2010; R. Zhao & Wang, 2011).
5. For example, one of the fundamental approaches is Scientometrics, which is used to gauge and analyze science (LoetLeydesdorff, 2001; Price, 1965).2
6.
7. One of the interesting and emerging areas in the field of Scienctometrics is the use of social network concepts for analyzing scientific knowledge (Hou, et al., 2008; Lee & Jeong, 2008; Nagpaul, 2002; Park, Hong, & Leydesdorff, 2005; Park & Leydesdorff, 2009; Pritchard, 1969; Wang, et al., 2010).3
8.
9.
10. However, in this article, we used social network analysis techniques (Wasserman & Faust, 1994) to visualize and gauge the core of scientific knowledge:
15. Can we visualize and model the underlying casual or theoretical relationship among theoretical constructs and models used in scientific literature by employing social network analysis techniques?6
18. Bridge—to determine bridging theories or constructs, etc.2) Conceptualize a research domain and derive the number of possible missing and potential links or researcher hypothesis graphically and mathematically (using directionless NCs). 3) Explore the strengths and limitations of a research domain from the structural characteristics perspective. Note: throughout the article we use IS research domain to demonstrate NC concept 7
25. Similarly, Dewan and Riggins (2005), constructed graphical view of digital divide research domain.
26. More recently, Khan et al., (2010a) proposed the shape of EG research taking place from developing and developed country perspective
27. Khan et al., (2010b) proposed mapping and visualizing e-government research theoretical constructs using mathematical and conceptual models to identify certain strengths and limitations, such as, identifying a missing links within a theoretical domain and a potential research hypothesis not visible otherwise. 9
43. A graphical view of the digital divide research (Dewan and Riggins, 2005), and
44. The shape of e-participation by Saebo et al. (2008) are good examples of non-casual conceptualization.Figure 2: Shape of the literature on e-government issues/topics (Khan et al., 2011) 13
51. NC of the Swar (2011) model Table 1. IS/IT out sourcing key constructs in terms of centrality measures. Table 2. IS/IT out sourcing domain network level properties. 16 Figure 4 NC of Swar (2011) model
54. There may be situations where a node (s) may be optional (or can be skipped) while constructing the NC. Again theory, casual relationship, researcher’s choice, or characteristics of research domain/sub-domain will determine the optional node (s). Similar conditions apply to the mandatory component (s). 17
55.
56. Based on the concept of direction in which one node can affect other, we can construct two types of NC networks. Let’s call them directional NC and direction-less NC.
58. In the direction-less NC, we are mainly interested in obtaining all possible ways (links) in which one node (s) can affect other in a research domain/sub-domain regardless of the theoretical or causal relationship among the nodes. In other words, in the direction-less NC, theoretical casual relationship among the node (s) is not considered.
60. The directional NC can only be constructed if a domain/sub-domain is well established and investigator has knowledge of all available theories and casual relationships (links) among nodes of a research domain/sub-domain under study.
61. All other types of NCs disused below can be either directional or direction-less in nature.18
62.
63. Direction-less NC may be applied in situations, for example, where researcher is interested in getting graphical view of a research domain which is new, or does not have enough theoretical background, or is not yet fully recognized discipline.
65. The primary purpose of directional NC, for example, can be to obtain a graphical view (or network) of a research domain/sub-domain which is well established and needs new nodes for expansion (e.g. interdisciplinary research); or
66. we are interested to model (graphically and mathematically) the relationship among nodes in a particular research domain/sub-domain; or
67. to identify the missing links; reveal hidden structures and characteristics of a research domain, for example, connectedness, centrality, density, etc 19
76. Let us assume that there are nnumber of societal factors “SF”, m number of organizational factors “OF”, and pnumber of technological factors “TF” that can affect EG adoption behavior.
77. So, in total we have n + m + p= N number of e-government adoption factors “EGAF” that can affect adoption behavior.possible combinations to choose from EGAF factors, where N= n + m + p possible combinations to choose from 2 SAs 23
78.
79.
80. Furthermore, for flexibility reasons, we can generalize the equation 2 so that it can accommodate different setups.
81. Let’s assume that there are M numbers of “EGAF” factors affecting EG adoption behavior, N number of levels for analyzing these factors, O stages of “EG” development, and P numbers of scopes available as shown in figure 2. Figure 6 Generalized form of NC in electronic government research context 25
82.
83. 27 Directionless NC to identify possible research areas Figure 7 Shape of the Literature on E-Government Issues/Topics (khan et al, 2010)
91. Can be written as: Solving Eq. 5 produced X = 146,475 (note that it is not density as well) unique ways of connecting the nodes. For simplicity reasons, Fig. 5 shows 32 (0.02%) of the possible ways of combining the nodes Fig. 6 shows 32 (0.02%) of the possible ways of combining the nodes. For example, we may be interested into investigate: the number of field studies which talk about social issues related to the ex-ante stage of EG from the perspective of developing countries or number of empirical studies which talk about organizational issues related to the ex-post stage of EG from developed countries perspective. ……Equation 5
92. Directionless NC to identify possible research areas Figure 8 NC of e-government research domain from adoption point of view 29
95. The units of analysis in casual conceptualization (or directional NCs) are specific representations of a research domain, for example, research models and constructs.
96. Thus, they can mainly be used to identify hidden structures that may reside within a complex research domain not visible otherwise using social networking concepts and techniques
98. non-casual conceptualization (or directionless NCs) can be employed to graphically model (i.e., produce a whole picture or layout) concepts and phenomena residing within a research domain/sub-domain and mathematically derive the number of missing and potential links or researcher hypothesesArticle 1 30
99.
100. Thus, the NC approach, particularly directionless NC, can only be applied to a scientific domain given that
106. For example, future research may construct a network of all the constructs and theories used in EG research and reveal its hidden structural characteristics which will help understand the structural differences among theories.
107. Other areas open for future studies are constructing domain level, sub-domain level, cross-domain level, and model level NCs for MIS research or social science research as a whole.31