Content area
The approach that knowledge can be divided into subordinate fields of discipline was already evident in ancient Greek philosophy. Currently interdisciplinarity is one of the basic measures for assessing research work. It was influenced by institutional changes and competitive in financing and research evaluation. The growing complexity of large projects has increased teamwork, combining the efforts of many scientists with experience in various fields. All modern attempts to organize and classify sciences as a starting point take the assumption of the unity of science as a specific type of cognition from which its varieties emerge. Therefore, an attempt was made at work to analysis of issues related to computer science in research papers in the field of economics is the main goal of the paper. The authors are going to analyse abstracts of research papers related to the field of economics and published by authors from 36 European countries and registered in Scopus database in the period 2011-2020. The main steps of the research are the following: preparation of a data set with abstracts of papers published by European authors in the field of economics in the period 2011-2020, identification of concepts appearing in abstracts and related to economics, identification of concepts appearing in abstracts and related to computer science, building a model of relationships between concepts belonging to these two areas of science, analysis of contribution of computer science issues in research papers from the field of economics. The ontology-based approach will be used for identification of concepts related to computer science and economics. The JEL classification system {Journal of Economics Literature, b.d.) and CSO ontology {Computer Science Ontology, b.d.) will be used for carrying out this step of the analysis. The description of relationships between computer science and economics issues will be based on bipartite graph model and ecological modelling. All analysis will be performed with the use of programs prepared by authors in R language.
Abstract: The approach that knowledge can be divided into subordinate fields of discipline was already evident in ancient Greek philosophy. Currently interdisciplinarity is one of the basic measures for assessing research work. It was influenced by institutional changes and competitive in financing and research evaluation. The growing complexity of large projects has increased teamwork, combining the efforts of many scientists with experience in various fields. All modern attempts to organize and classify sciences as a starting point take the assumption of the unity of science as a specific type of cognition from which its varieties emerge. Therefore, an attempt was made at work to analysis of issues related to computer science in research papers in the field of economics is the main goal of the paper. The authors are going to analyse abstracts of research papers related to the field of economics and published by authors from 36 European countries and registered in Scopus database in the period 2011-2020. The main steps of the research are the following: preparation of a data set with abstracts of papers published by European authors in the field of economics in the period 2011-2020, identification of concepts appearing in abstracts and related to economics, identification of concepts appearing in abstracts and related to computer science, building a model of relationships between concepts belonging to these two areas of science, analysis of contribution of computer science issues in research papers from the field of economics. The ontology-based approach will be used for identification of concepts related to computer science and economics. The JEL classification system {Journal of Economics Literature, b.d.) and CSO ontology {Computer Science Ontology, b.d.) will be used for carrying out this step of the analysis. The description of relationships between computer science and economics issues will be based on bipartite graph model and ecological modelling. All analysis will be performed with the use of programs prepared by authors in R language.
Keywords: interdisciplinarity, computer science, economics, scientific papers, bipartite graph
(ProQuest: ... denotes formulae omitted.)
1.Introduction
The pursuit of ever greater specialization and at the same time internal coherence is the basis for the development of science. This development gained particular momentum in the XIX century, leading to the identification of a large number of detailed disciplines on the one hand, and on the other hand, showed the bonds occurring in various planes between disciplines previously separated as independent. This was due to the discovery of common aspects for various phenomena, the dissemination of mathematical methods used in various fields of science {natural, social). The manifestation of this process is the increase in interdisciplinary links.
The concept of interdisciplinarity is often raised in the literature on the subject. Interdisciplinarity is define as direct or indirect use of knowledge, methods, techniques, devices {or other products) as a result of scientific and technological activities in other fields (Tijssen, 1992). We can define interdisciplinary research as any study or group of studies undertaken by scientists from two or more distinct scientific disciplines. The research is based upon a conceptual model that links or integrates theoretical frameworks from those disciplines, uses study design and methodology that is not limited to any one field, and requires the use of perspectives and skills of the involved disciplines throughout multiple phases of the research process (Aboelela et al., 2007). We can also find definition that describe interdisciplinary research as a mode of research by teams or individuals that integrates from two or more bodies of specialized knowledge or research practice (Porter et al., 2006):
* perspectives/concepts/theories and/or
* tools/techniques and/or
* information/data.
The success of interdisciplinary research depends on many factors. Among them we can mention (Aboelela et al., 2007):
* Institutional factors - an explicit institutional commitment to interdisciplinarity and sufficient resources.
* Team factors - communication, leadership, and trust.
* Individual characteristics of team members - commitment, flexibility, and being agreeable to work with.
In the literature on the subject, we can find various motives for scientists introducing interdisciplinary research:
* Many researchers have conducted interdisciplinary research because they have recognized the limitations of their disciplinary perspective (Aboelela et al., 2007).
* There is a need for innovative solutions that combine knowledge from various scientific disciplines (Rijnsoever & Hessels, 2011).
* Interdisciplinary research has a positive influence on knowledge production and innovation (Rijnsoever & Hessels, 2011).
* Scientists expected future benefits in terms of publications, recognition and funding (Rijnsoever & Hessels, 2011).
In this article, the authors intend to analyse the interpenetration of issues related to computer science in economics. Identified ontology-based topics and models of bipartite graphs will be used to study this find.
2.Interaction of economic and computer sciences
Currently, an interdisciplinary approach to the study of business processes can be seen. Economics is most often treated as social science, although there is no shortage of researchers who make extensive use of advanced mathematics - science.
Since economics is precisely a social science and borders on other important academic disciplines, such as: political science, psychology, anthropology, sociology, philosophy or history, research on the interdisciplinary nature of economic sciences is most often conducted in this area. It seems to be missing work looking for the influence of computer science in economic sciences.
Computer scientists have a long history of developing tools useful for advancing knowledge and practice in other disciplines. Big data sets and the digital revolution are changing the world we live in, influencing decision making and policy making by individuals, corporations, institutions and society as a whole. Over the past 20 years, the development of the internet has increased the interaction between economic sciences and computer science and these interaction influenced economic theory in the following ways (Blume et al., 2015):
* New scientific problems - e-commerce, new applications including network management, online social networks.
* New issues in areas already popular in economics e-commerce, decision theory, market design, network.
* New methods to existing problems.
Literature on the subject gives many links between economics and computer science. Among them we can point to macroscopic insights about the economic incentives of Internet users and their service providers to very specific game-theoretic models for the behaviour of low-level Internet protocols for basic functionality, such as packet routing. Across this entire range, the economic insights often suggest potential solutions to difficult problems (Kearns, 2005).
The interaction of economics and IT in the context of industry 4.0 is becoming particularly important. Three industrial revolutions were related to mechanization, electricity and information technologies in human production. The fourth industrial revolution now involves computer generated product design and three dimensional (3D) printing, which can create solids object by building up successive layers of materials. Now a Fourth Industrial Revolution is building on the Third, the digital revolution that has been occurring since the middle of the last century (Xu et al., 2018). The concept of Industry 4.0 entails necessary changes in the operational processes of companies and forces changes in doing business in the field of relations with clients, work environment, production, technology etc.
With such connections, the question arises of how to build classification schemes of research disciplines and fields, to reflect the nature of modern science, and how to reconcile the hierarchical nature of classification with links in science, which are often bilateral. This article will discuss two classifications Journal of Economics Literature - JEL and Computer Science Ontology - CSO. First classification the JEL (Journal of Economics Literature, b.d.) is system originated with the Journal of Economic Literature and is a standard method of classifying scholarly literature in the field of economics. It is used in many of the AEA's published research materials. Second classification the CSO (Computer Science Ontology, b.d.) is a large-scale ontology of research areas that was automatically generated using the Klink-2 algorithm on the Rexplore dataset, which consists of about 16 million publications, mainly in the field of Computer Science.
3.Research methodology
3.1 Research scope and goals
The research presented here is focused on the analysis of interdisciplinarity in the field of economics, in particular to the analysis of the participation of computer science related issues in scientific papers on economics.
The authors defined the following detailed objectives:
* the analysis of abstracts of research papers from the area of economics to identify topics related to main subareas existing in economics and computer science area;
* the analysis of relationships between subareas defined within economics and computer science areas in the light of content analysis of abstracts.
3.2 Identification of topics occurring in abstracts and related to main subareas of economics and computer science
Identification of topics occurring in abstracts and related to main subareas of economics and computer science was performed by the analyser presented in Figure 1.
A collection of abstract plays a role of input data and a matrix showing a contribution of every subarea of a given discipline in every document is produced as a result of the analysis.
The analyser has an ontology-based character. The ontology describes a given scientific discipline and has a form of a directed graph in which nodes represent concepts existing in the discipline and edges represent ties between them (edges lead from more to less general concepts).
To every concept in the ontology a list of patterns should be assigned. One pattern describes one phrase which can be used in an abstract to represent a given concept. Patterns are defined using a notation proposed in (Kovaleva at al., 2020) indicating which words are obligatory and which have optional character or form a sequence of words (key-phrases).
The analyser processes every document from the collection of abstracts separately. The process of analysis is presented in Figure 2.
Every document, using a sliding window technique, is transformed into a sequence of phrases. Next, iteratively, for every pattern and every phrase, two coefficients of similarity are calculated. The first for original pattern and original phrase, and the second one - for lemmatized versions of them. The maximum calculated for these two values represents the similarity between a given phrase and a given pattern. Taking the maximum for all phrases from a given document and a given pattern, a pattern's contribution in a document can be estimated. The results form a document - pattern matrix.
During the last step of the analysis, the content of the document - pattern matrix should be transformed into document - concept matrix.
For every main subarea, the number of patterns related to this area and recognized in a given document was calculated. These values form a document - main subareas matrix. Finally all positive values in the S matrix were transformed into ones. Elements of the S matrix has very intuitive interpretation. If s¿;- = 1 then the y-th subarea is represented in the i-th document, otherwise (is s¿;- = 0) the y-th subarea is not mentioned in the i-th document.
3.3The analysis of relationships between main subareas defined in two scientific disciplines
Let's consider a collection of abstracts A and two scientific disciplines X and Y.
For the discipline X the set of main subareas has a form:
...
Similarly, for the discipline Y, the set of main subareas has a form:
...
Assuming that and Sy describe the contribution of main subareas of discipline X and discipline Y in every document from the A collection, the bipartite graph G can be defined by defining its adjacency matrix B:
...
in which an element bu can be interpreted as a number of abstracts from the collection A in which simultaneously main subareas mx¿ and mYj were recognized.
Three aspects of the G graph can be studied:
* subarea strength
* subarea specificity
* graph model modularity
3.3.1 Subarea strength
Separately for every discipline, the analysis of the strength of every main subarea was performed. The significance of every subarea was expressed with the use of two measures: node's degree and node's strength.
The degree value for a node representing a given subarea indicates the number of subareas from the second discipline which are connected with the given node.
The degree value for nodes representing subareas from the X discipline is defined as:
...
and for the Y discipline it has an analogous form:
...
Dividing above values by the maximum (equal to the number of subareas in the opposite discipline), the normalized version of a degree measure can be formulated:
...
and
...
The normalized version always belongs to a range [0; +1] and therefore is more convenient for interpretation.
According to (Bascompte et al., 2006), a node's strength can be measured as a sum of connections with nodes from the second group. Using this idea, the strength of main subareas form the X discipline can be expressed as:
...
and the strength of main subareas from the Y discipline may be calculated with the use of the following formula:
...
To make the interpretation easier, the normalized version of above two measures can be used which are normalized to the [0; +1] range. They can be calculated for main subareas from the X discipline:
...
and, in the similar way, for main subareas from the Y discipline:
...
3.3.2 Subarea specificity
Subarea specificity reflects the diversity of interactions between a given subarea from the first discipline and subareas from the second discipline. Low diversity can be interpreted as low specificity, whereas high diversity indicates high specificity. Referring to Poisot at al. (2012), the measures of subareas specificity for the X discipline can be defined as:
...
and for main subareas from the Y main subareas:
...
where:
...
and:
...
In addition to specificity measure for every subarea, the specialization index H2 can be calculated (Blüthgen at al., 2006).
The specialization index H2 can be treated as an aggregated measure informs about the subareas specificity. It is based on Shannon information theory. For calculation the Hļ index, all elements of the В matrix should be transformed into probabilities:
...
Next, the two dimensional Shannon entropy can be calculated:
...
The specialization index H2 is defined as a normalized version of the H2 index:
...
where is the maximum and H™ln is the minimum value of the H2 index calculated for the given matrix B. The H'2 coefficient always belongs to the range [0; +1]. It is close to zero for low specialization (subareas from the first discipline interact with many subareas from the second discipline) and it is close to one when every subarea from the first discipline is connected with very limited number of subareas from the second discipline.
3.3.3 Graph model modularity
Modularity analysis allows to check if connections between nodes from two groups are evenly distributed. If the number of connections between pairs of nodes is diversified, then compartments (subgraphs completely separated) or communities (subgraphs which are connected with other parts of a graph, but their ties with other subgraphs are notably weaker then ties existing within them). While identification of compartments is relatively simple, finding communities can raise some problems. The most popular approach of community detection is based on maximization of modularity measure presented in (Dormann et al., 2014).
Assuming that a community is formed by some subareas taken from the discipline X and some strongly connected with them some subareas from the discipline Y, the function 5 ^c(mx i), c(myj·)j equal to 1 if subareas mxi and mY j belong to the same community or equal to 0 otherwise can be defined. Then the modularity measure can be stated as:
...
where:
...
and:
...
To solve the problem stated above, subareas from disciplines X and Y should be assigned to communities in the way which maximize the Q value.
4.The analysis of the contribution of computer science issues in research papers published in the field of economics
The analysis was performed with the use of abstracts of research papers published in the field of economics, prepared by authors from 36 European countries and registered in the Scopus database from 2011 to 2020 year. The total number of documents taken into account in the analysis was 124460.
For the description of the area of economics the JEL ontology (Journal of Economics Literature, b.d.) was used. The computer science area was defined by the CSO ontology (Computer Science Ontology, b.d.).
The distribution of documents over the main subareas from the JEL ontology is presented in Figure 3.
For the same set of abstracts, the distribution of documents over main subareas fromtheCSO ontology is presented in Figure 3.
The bipartite graph showing connections between main subareas from economics and computer science discipline is presented in Figure 4.
Subarea's strength is represented by a width of a block representing a given concept. And the width of an edge indicates the strength of a connection.
The most popular connections between subareas from economics and computer science disciplines are presented in the Table 1.
Specificity indexes for subareas are presented in the Table 2.
And the HÓ index is very small and equal to 0.000519.
Specificity indexes and Я2 index indicate that for every subarea from the economics discipline, the distribution of connections to subareas from the computer science discipline is almost identical and vice versa.
5.Conclusions
In the paper the analysis of the contribution of topics related to computer science area to economics was presented. It seems that the fusion of ontology-based topic identification and bipartite graph models form a useful tool for this task.
The results show that concepts related to computer system, Internet and artificial intelligence have huge impact on research topics considered in economics area. The analysis of subareas' specificity shows that its level is rather low. Also it is worth to underline that during the research process four communities were identified.
Acknowledgements
The research has been carried out as part of a research initiative financed by the Ministry of Science and Higher Education within "Regional Initiative of Excellence" Programme for 2019-2022. Project no.: 021/RID/2018/19. Total financing: 11 897 131.40 PLN
References
Aboelela, S. W., Larson, E., Bakken, S., Carrasquillo, O., Formicola, A., Glied, S. A., Janet, H., & Gebbie, M. K. (2007) "Defining Interdisciplinary Research: Conclusions from a Critical Review of the Literature", HSR: Health Services Research, Vol 42, No. 1, pp 329-346.
Bascompte, J., Jordano, P. & Olesen, J.M. (2006) "Asymmetric coevolutionary networks facilitate biodiversity maintenance", Science, No. 312, pp 431-433.
Blume, L., Easley, D., Kleinberg, J., Kleinberg, R., & Tardos, É. (2015), "Introduction to computer science and economic theory", Journal of Economic Theory, No. 156, pp 1-13.
Blüthgen, N., Menzel, F. and Blüthgen, N. (2006) "Measuring specialization in species interaction networks", BMC Ecology No. 6, pp 12.
Dormann, C.F., Strauss, R. (2014) "A method for detecting modules in quantitative bipartite networks", Methods Ecol Evol, No. 5, pp 90-98. doi:10.1111/2041-210X.12139. https://www.aeaweb.org/econlit/ielCodes.php?view=iel https://cso.kmi.open.ac.uk/faq
Kearns, M. (2005) "Economics, Computer Science, and Policy", Issuses in Science and Technology, Vol 21, No.2.
Kovaleva, A.S., Lula, P., Tuchowski, J. (2020) "Ontology-Based Measurement of Study Program Innovativeness in the Area of Eco-nomics and Management". In: Babić, V., Nedelko, Z., and Hershey, P.A. (eds.) Handbook of Research on Enhancing Innovation in Higher Education Institutions. pp 381-407. IGI Global, Hershey (2020)
Poisot, T., Canard, E., Mouquet, N., & Hochberg, M.E. (2012) "A comparative study of ecological specialization estimators. Methods in Ecology and Evolution", No. 3, pp 537-544, https://doi: 10.1111/j.2041-210X.2011.00174.x.
Porter, A. L., Roessner, J. D., Cohen, A. S., & Perreault, M. (2006) "Interdisciplinary research: meaning, metrics and nurture", Interdisciplinary Research: Meaning, Metrics and Nurture, Vol 15, No. 3, pp 187-195.
Rijnsoever, F. J. van, & Hessels, L. K. (2011) "Factors associated with disciplinary and interdisciplinary research collaboration", Research Policy, Vol 40, No. 3, pp 463-472.
Tijssen, R. J. W. (1992). "A quantitative assessment of interdisciplinary structures in science and technology: coclassification analysis of energy research", Research Policy, No. 21, pp 27-44.
Xu, M., David, J. M., Kim, S. H. (2018) "The Fourth Industrial Revolution: Opportunities and Challenges. International Journal of Financial Research, Vol 9, No. 2, pp 90-95.
Copyright Academic Conferences International Limited Oct 2020