Analytical assessment process of e-learning domain research between 1980 and 2014

Applying some methods to reduce the time and expenditures of training is inevitable in existing circumstances. Many educational organisations have realised the importance of Electronic Learning (E-learning) and tried to use this approach in leveraging their academic classes. As research in e-learning domain has become one of the most important and interesting subjects, observation in emerging and fading trends of e-learning is a need for the scholars and industry professionals who are interested to study and work in the field. This paper has triggered the investigation and depicting of scientific trends in e-learning by using two scientometric methods named burst detection and clustering analysis. By applying two mentioned methods, the hot topics were identified in the field of e-learning.


Introduction
Although modern methods of teaching would never replace with traditional methods, presenting courses in a mere traditional way would not be completed unless along with some new technologies; as it facilitates knowledge transfer (Furió et al., 2015).Modern technologies and globalisation have created dramatic changes in all aspects of human life.Development in new knowledge and skills helps to apply information.Therefore, new knowledge of "quick and available tutorial" leads to new types of training (Milošević).In recent years, E-learning has experienced an increasing trend as an acceptable solution for effective and swift learning.E-learning has been defined as the use of internet technology to enhance the quality and quantity of learning (Mayes and de Freitas, 2004).E-learning has been defined as a part of educational process at different levels of education from primary to higher education as well and can even be used in corporation environments in order to integrate entire staffs and reduce the time and expenses of training (Welsh et al., 2003).The benefits of this educational system increase the number of training programs in different fields (medical, management, engineering, medicine, information technology, electronics and telecommunications) (Stänescu and Muçat, 2015).On the other hand, all high expenditure of E-learning software, high initial capital and a lot of time to enter the market, lead to emergence of some approaches that attempt to reduce the disadvantages of e-earning through the combination of E-learning into other technologies.Among all these technologies cloud computing (Kihara and Gichoya, 2014) and using mobile phone (Joo et al., 2014) have been used more than others to train.That is why research trends identification is a great help for researchers in this field to recognise emerging trends and choose the subject of future research as time goes on.In the following text it has been tried to review and analyse the published articles in the field of electronic learning in the most prestigious academic database in the world, Web of Science (WoS), by two methods called burst detection and clustering analysis to identify the hot topics in E-learning domain.
Chen and Lien (2011) used an author co-citation analysis (ACA) which is an analytical method for identifying the intellectual structure of specific knowledge domains through the relationship between two similar authors, analysed intellectual structure of E-learning from the perspective of management information systems (MIS).Lin and Hu (2015) 2014) identified six research themes as result in the E-learning, which categorised into four dimensions: (1) integration of knowledge management with E-learning, (2) E-learning for continuing education and professional development, (3) use of social media for Elearning and (4) E-learning in the healthcare sector.Kalmykova et al. (2016) formed the model of information streams for the purpose of the choice of an optimal variant of network interaction when forming educational trajectories is offered.Aparicio et al. (2014) identified 22 related E-learning terms used in literature and organised these concepts in a chronological way and then identified new concepts trends in E-learning and compared their publication growth rate with E-learning growth rate from 2010 to 2013.
Table 1 summarises studies have used this method in E-learning process.Generally, the application of bibliometrics methods can be seen conspicuously in so many disciplines and subject areas like e-business (Jalali and Mahdizadeh, 2016;Jalali and Park, 2017;Vanani and Jalali, 2017), informetics (Bar-Ilan, 2000), Thesaurus construction (Lykke Nielsen, 2001), E-Government (Jafar Jalali, 2016), and health information (Garcia-Lacalle et al., 2011;Bowler et al., 2011).Hasanagas et al. ( 2010) developed E-learning system in environmental science management is a challenging task in the area of forest and general rural development policy.They discussed the development of a GIS-based model which includes region-based scientometrics, regarding policy field communication by using VISONE software to recognise the most "important" in a network and is going to be applicable in E-learning for various target groups (e.g.students who are specialised in forest policy and rural policy analysis, lobbyists, policy makers).Breznik (2016) used the social analysis to identify the most important research institutions in Slovenia, and reveal clusters of research organisations which collaborate.
In this article scientometrics method is used to assess investigations that has been done in this field.Scientometrics is an authentic research method that can be used to depict part of the information (Hook and Börner, 2005).This algorithm is a quantitative method which can be used for an extensive study of academic publications (Chen et al., 2011).Scientometrics objectives include: Providing a pictorial and graphical output to create a broad perspective in specific domain as well as structural details of this scope and its outstanding features by scrutinising a large number of papers (Hook and Börner, 2005).Considering too amount of information available in the scientific sphere, providing picture of changes trend in academic disciplines enhance the information communication (Börner, 2012).This article seeks to examine the issue of how E-learning course has changed between 1980 and 2014.To analyse the trends Burst Detection algorithm is used and data based on keywords, titles is categorised separately and both most and least are discussed.This study is used the VOSViewer (Van Eck and Waltman, 2009) and Sci2 (Team, 2009) applications.In this regard, upon presentation of clustering and burst detection algorithm, the results of the analysis are presented during the period.

Methodology
After accessing to Web of Science (WoS) core collection database, we have extracted the related papers to the field of E-learning.Next, we applied the Burst detection as well as clustering algorithm.We visualised both algorithms using visualisation techniques by applying on the keywords and titles of E-learning papers.In the results section, we measured and visualised the outputs of the burst detection and clustering algorithms.The steps of the methodology that has been used in the paper is in Figure 1.
Figure 1 The methodology of current study

Data gathering
The data used in this study, has been gathered from Web of Science (WoS).The word "E-learning" was applied on WoS search engine between 1980 and 2014 and 13,895 articles were retrieved.A summary of the collected data is presented in the Table 2.
What is observed from Table 2, indicating that E-learning has experienced a significant increase in 2007 and 2008 and in the years leading up to 2014 has taken a downward trend.Current research has experienced the following methodology in Figure 1.

Burst detection
The Burst Detection algorithm has been used for detecting the scientific emerging trends (Guo and Borner, 2011;Swar and Khan, 2014), and was first used in 2003 by Kleinberg (2003).Keywords and titles of scientific papers are the important components of scientific papers that can express the theme of scientific trends.It should be noted that because of not adding any values to output results, we did not consider the term "Elearning" to neither of the steps of the analysis section.The results of burst detection algorithms are applied on the field of E-learning in the network of 20 keywords and titles as it is shown in Tables 3 and 4, respectively.
Table 3 listed all of the top 20 emerging issues, sorted by their weights with respect to keywords.As it is indicated, the word web includes the highest repetition and subsequently has enjoyed the most weight among other keywords.Other top keywords are object, cloud, leam, scorm, xml, grid and metadata, addressing the important technologies and methods for constructing fundamentals of E-learning platforms.Besides, by taking a closer look at other top keywords, we can conclude that new technologies such as cloud computing (understood from the term cloud) is in an upswing technology in the field of E-learning.The other keywords such as standard, smart, media and multimedia paid to important characteristics of E-learning domain.For instance, multimedia has been effectively utilised in E-learning and its advantages for classroom learning are well designed.Figure 2 shows graphical view of the output of burst detection algorithm on the E-learning keywords, obtained from Table 3.The horizontal axis indicates the time span of keywords and the vertical axis indicates the weights of keywords.This figure also indicates that start of important E-learning articles is from the year 2000 by looking at the keywords of the articles.Also, in Table 4, the burst detection algorithm has been performed on the network of the titles of the E-learning articles.As Table 4 indicates, cloud technologies have a remarkable place among other titles of Elearning articles.We can also infer that the terms such as web, grid, agent, metadata, servic, distribut and semant enjoyed the most weights among other titles.For instance, these terms show the importance of technological advances and applications of semantic web, web 2.0 and Distribution systems in E-learning area.The figure above also shows words such as metadata, agent, librari and interface have lasted for seven, five, four and two years respectively.On the other hand, the terms such as, cloud, social, online and trial are the hot topics because they ended in the year 2014.

Clustering
Clustering analysis is a powerful and useful tool for analysis of information flow (Marshakova, 1973).Clustering Method is used for searching scientific literature, scientific issues, identifying leading researchers and classic papers, the birth of a new scientific process, track the evolution of each discipline, recognising the "points of growth", analysis and improvement of scientific activities in scientific historical research communities.We applied clustering algorithm on the network of keywords and titles, indicated in Table 5 and Table 6, respectively.Besides, the clustering network of keywords and titles of E-learning articles have been visualised in the Figures 4 and 5, respectively.A summary of the results of clustered keywords has been shown in the Table 5.We indicated the most significant clusters and their related keywords in the Table 5. Figure 4 shows a schematic view of Table 5.It should be noted that the keywords with the same colour are considered in a same cluster.Cluster 1 covers the subjects such as Education, information, knowledge, training; which are the basic and fundamental subjects in E-learning area.Cluster 2 devotes to behavioural aspects of E-learning area.Cluster 2 is devoted to learning approaches in Elearning domain such as text mining and web mining.Cluster 4 indicates significant tools for providing the structure of E-learning frameworks.
We also applied clustering techniques to titles of E-learning articles which top frequent keywords have been clustered in Table 6.For showing the important clusters, we selected 15 top terms.Cluster 1 indicates the learning process and the technologies for constructing a successful E-learning platform.
Cluster 2 indicates the basic platforms and services such as virtual classes and E-learning applications.Cluster 3 contains the words such as taxonomy which is a base for providing the online learning classes.The terms such as semantic web and learning contents in cluster 4 represents the importance of these terms in intelligent E-learning systems.
We also visualised the whole clustered network of titles of E-learning domain in Figure 5.

Conclusion
In this study, we applied scinetometrics methods such as clustering and burst detection over titles and keywords of E-learning domain.The results of burst detection algorithm, represent the emergence and disappearance themes through academic studies in the field of E-learning.By applying clustering analysis, it was inferred that the main current research directions in the field and by segmenting them into number of clusters, the important hot topics were identified.The findings indicated the current research direction for the scholars who were willing to understand the emerging and fading themes in the field of E-learning.The results also make evidence using remote access to E-learning frameworks.Web of Science database were used in this study.It is suggested that in the future research, more data from several databases such as Scopus and ProQuest databases have been studied and the results were compared with the current study.

Figure 2
Figure 2 The diagram of top 20 terms in E-learning by burst detection on keywords

Figure 3
illustrates the Table 4 by applying title terms of E-learning in a time-span from 2001 to 2015.This figure also indicates that beginning of important E-learning articles is from the year 2001.

Figure 3
Figure 3 The diagram of top 20 terms in E-learning by burst detection on titles

Figure 4
Figure 4 Clustering based on keywords

Figure 5
Figure 5 Clustering based on titles

aimed to provide visualisation of trends and research fronts in E-learning research from 2002 to 2013 through selecting five core journals from SSCI (British Journal of
Teaching International and Educational Technology Research and Development) and two conferences (Educational Multimedia, Hypermedia & Telecommunications and IEEE International Conference on Advanced Learning Technologies).Cheng et al. (

Table 1
Previous studies of E-learning

Table 2
Year of publication, articles number and percentage of E-learning term in the WoS

Table 3
First 20 obtained terms of burst detection on keywords

Table 4
First 20 obtained terms of burst detection on keywords

Table 5
Clustering table frequent keywords Among 23 top words, we clustered them into four groups.Words such as Education at the first cluster, behaviour in the second cluster, machine learning, neural network, text mining in the third cluster and cloud, remote learning in the fourth cluster are recognised as outstanding keywords among other keywords .

Table 6
Clustering table frequent titles