Intellectual structure of knowledge in iMetrics: A co-word analysis

Highlights

• • In order to build a complete structure of subjects areas in iMetrics, both types of keywords are included in this study.

• • Application of hierarchical clustering analysis led to the formation of 11 subject clusters in iMetrics.

• • Analysis of the strategic diagram showed two most comprehensive themes of iMetrics.

• • In spite of their similarity with those of Courtial (1994), clusters in this study were different from several aspects.

Abstract

As an iMetrics technique, co-word analysis is used to describe the status of various subject areas, however, iMetrics itself is not examined by a co-word analysis. For the purpose of using co-word analysis, this study tries to investigate the intellectual structure of iMetrics during the period of 1978 to 2014. The research data are retrieved from two core journals on iMetrics research (Scientometrics, and Journal of Informetrics) and relevant articles in six journals publishing iMetrics studies. Application of hierarchical clustering led to the formation of 11 clusters representing the intellectual structure of iMetrics, including “Scientometric Databases and Indicators,” “Citation Analysis,” “Sociology of Science,” “Issues Related to Rankings of Universities, Journals, etc.,” “Information Visualization and Retrieval,” “Mapping Intellectual Structure of Science,” “Webometrics,” “Industry–University–Government Relations,” “Technometrics (Innovation and Patents), “Scientific Collaboration in Universities”, and “Basics of Network Analysis.” Furthermore, a two-dimensional map and a strategic diagram are drawn to clarify the structure, maturity, and cohesion of clusters.

Introduction

Metric studies have been developed as a subsidiary branch of Library and Information Science over time. Various concepts of the field, such as bibliometrics, scientometrics, informetrics, webometrics, and technometrics are found in LIS journals. As proposed by Milojevic and Leydesdorff (2013), these concepts have similar goals and methods, and can be grouped under a research subset titled information Metrics or iMetrics. Using co-citation, bibliographic coupling, and co-word methods for exploring research topics in LIS, Chang, Huang, and Lin (2015) found that iMetrics was the most significant topic in LIS subsets. As an independent trend, iMetrics is not only emerging, but is also evolving into its socio-cognitive nature (Milojevic & Leydesdorff, 2013). By the application of common techniques in iMetrics, one can collect and evaluate data onto research trends and researcher status in different disciplines while evaluating research output concurrently (Hunter, 2009, Stidham et al., 2012, Webster, 2011, Weightman and Butler, 2012, Zyoud et al., 2014). Due to its applications, iMetrics is also employed by researchers in other disciplines.

Considering the gradual emergence and development of iMetrics, a comprehensive macro image of research on iMetrics should be drawn, and its scientific development needs to be explored, in order to enquire into its advancement in a temporal continuum. One of the techniques employed for analyzing the knowledge structure of diverse fields is studying the relation between words used in various parts of a document, including the title, abstract, keywords, etc. This technique is called “co-word” analysis, and is a well-established and effective approach, that can show the intellectual structure of a research field (Ronda-Pupo & Guerras-Martin, 2012). It is an approach used for establishing a subject similarity between two documents (Rokaya, Atlam, Fuketa, Dorji, & Aoe, 2008). Co-word analysis presumes that a group of aggregated keywords could indicate underlying themes, and that co-occurrences of keywords could show the associations with the underlying themes (Hu & Zhang, 2015). By employing co-word analysis, one can determine the major topics in a field, in addition to its semantic structure and evolution over the time. In co-word analysis, it is supposed that frequent words have more meaning of an effect on a field than the less frequent ones. It help in determining both the emerging and the developed subject clusters to suggest the research path in the future (Lee & Su, 2010).

The frequency of word occurrence is a principal measure in content analysis. This measure is used for exploring the major topics in a research field by giving attention to highly frequent words. In other words, the frequency of a given word is an indicator of the importance of the word and its notion. Keywords have the potential for effectively describing the contents of a paper. If two keywords occur simultaneously in a paper, they have a semantic relationship (co-word/co-occurrence). The higher co-occurrence frequency of two keywords implies the more correlative they are (Liu, Hu, & Wang, 2012).

Like other co-occurrence analyses, particularly that of co-citation, co-word analysis is one of the fundamental methods for demonstrating the relationship among concepts. It is used to determine research frontiers in academic disciplines and explore knowledge structures in various research fields (Hu et al., 2013, Ravikumar et al., 2015, Stegmann and Grohmann, 2003, Xie, 2015). By studying and analyzing the co-occurrence of keywords in the papers of a certain field, one can draw an instant picture of interesting topics within the field (Ding, Chowdhury, & Foo, 2001). In other words, there are collections of concepts in each scientific and technological field that build its knowledge structure. These concepts are expressed as keywords that are made for describing and naming them. Exploring concepts and the relationship between them by means of word relations in documents eases the creation of a scientific map.

As stated earlier, co-word analysis is one of the commonest approaches to iMetrics which allows us to reveal the emerging thematic clusters and the changes of traditional thematic clusters in order to forecast the path of coming researches (Lee & Su, 2010), and to study it's conceptual and semantic relations (Leydesdorff & Welbers, 2011). In addition, the intellectual structure of scientific domains can be examined as forming a cluster via clustering techniques and multidimensional scaling (Cho, 2014, Yan et al., 2015). The data relating to co-word analysis as well as the data relating to other co-occurrence analyses (such as co-citation and co-authorship) have the potential of being analyzed using multidimensional scaling, network and cluster analysis and to show the structure of knowledge in a given field (Allendoerfer, 2008).

Finally, the use of novel technologies in network analysis can reveal the ruling relationships in co-word analysis and deeply examine these complex relationships and depict the structure of knowledge in a specific field. Studying the knowledge structure can be fruitful for both researchers and science policymakers. Although co-word analysis is a kind of iMetrics technique, iMetrics itself is not examined through co-word analysis using relatively complete records. For this purpose, i.e. using co-word analysis, this study aims at investigating the intellectual structure of iMetrics during the period from 1978 to 2014. This paper tries to answer the following questions:

• 1. Can the intellectual structure of iMetrics be visualized and represented using hierarchical clustering?

• 2. Can the intellectual structure of iMetrics be visualized and represented using multidimensional scaling?

• 3. How are topics and clusters of iMetrics represented by the strategic diagram in terms of maturity and development?