**2. Methods**

One of the world's largest databases of scientific literature is Elsevier's Scopus, which contains approximately 18,000 titles from more than 5000 international publishers, including coverage of 16,500 peer-reviewed journals in the areas of Science, Technology, Medicine, and Social Sciences, including the arts and humanities. This is the methodological basis of this study, which has been used successfully in other bibliometric studies [34].

The coexistence of two large scientific databases, Scopus and Web of Science (WoS), raises the question of the stability of the statistics obtained by one or the other sources of information. Several studies have measured the overlap between databases and the impact of using di fferent data sources for specific research fields on bibliometric indicators, demonstrating a larger number of journals indexed by Scopus compared to WoS [35]. Regarding the overlap, 84% of the WoS titles are also indexed in Scopus, while only 54% of the Scopus titles are indexed in WoS [36]. For example, some studies related to citations in the papers conclude that each database covered 90% of the citations in the other database when the citation period is limited to Scopus citation coverage for 1996 and beyond [37].

The methodology followed in this work is described in Figure 2. The search term is first consulted in the Scopus database (1). The term smart contract was used for the entire historical series up to 2018, where the exact search query was TITLE-ABS-KEY (smart AND contract). The resultant search is exported to Comma Separated Value (CSV) text format (2) for each of the fields studied (i.e., publications by year, type of publication, publications by category indexed by Scopus, publications by country, publications by institutions, and keywords and their frequency). Thirdly (3), the previously downloaded text files are imported into Excel, and all of them are represented without removing any data. Fourthly (4), the keywords are represented by means of the free online software Word Art (https://wordart.com/) to obtain a cloud words where the most important ones are highlighted. Fifth (5), the methodology was developed to analyze the scientific communities or clusters associated with this thematic. The exported information of the complete search was imported in csv format in the free and online bibliometric analysis software called VOSviewer (http://www.vosviewer.com/). Here, the relations between the countries, interpreted through the co-authors of every one of the works, were analyzed, and the research clusters of the works were analyzed, using the relations between all keywords of all the works.

**Figure 2.** Methodology chart.

With respect to the chosen software, it should be noted that for the direct representation of results, bar charts, percentage distribution, or lines of evolution over time, a spreadsheet has been used via Microsoft Excel, which allows the direct import of the csv format exported by the Scopus database. For the clouds of words, the Word Art software has been chosen because it is free and online and allows the import of data from excel. Finally, the community detection software, we also opted for free software available online that allows the direct import of data in csv format exported from Scopus.

Finally, the community detection software used was the VOSviewer, which was also chosen for being free software available online that allows the direct import of data in the csv format exported from

Scopus and also allows the figures to be exported to a large range of graphical formats. The VOSviewer delivers three displays: Network visualization as clusters, overlay visualization as temporal evolution, and density visualization. In all cases, the parameters chosen for the analysis were: normalization method (association strength), layout (attraction 2, repulsion 0), clustering (resolution 1.00, minimum cluster size 1), and rotate (90 degrees).

#### **3. Results and Discussion**

#### *3.1. Evolution of Scientific Production, Languages, and Types of Documents*

The search yielded just over 1700 documents up to the year 2018, the evolution of which is reflected in Figure 3. This result shows that the first works published date back to the 1980s, but it was not until 2003 that they began to reach a certain relevance, with the volume of publications stabilizing at 50 works per year. However, the most remarkable increase can be seen from 2014 onwards, at which point the upward trend can be considered exponential.

**Figure 3.** Evolution of scientific publications related to smart contracts.

If the results are analyzed according to the type of publication (Figure 4), it is observed that most results are communications presented at congresses (53%; 46% conference papers and 7% conference reviews), followed by articles in journals (38%; 33% articles and 5% reviews), as well as books and book chapters (2%). The rest is distributed among other formats such as editorials, notes, and letters. The high percentage of conferences in relative terms is because it is a very recent technology or area of interest. Thus, when the subject matter of research is consolidated, the percentage of books is higher and, above all, the number of articles in relation to number of congresses is higher. Note that scientific conferences are usually on very specific scientific topics and are aimed at sharing ideas among researchers. In short, these scientific events are key activities for the process of knowledge dissemination, for the presentation of new findings, and for the development of science in a community.

**Figure 4.** Distribution of publication type in relation to smart contracts.

These papers are written mostly in English, accounting for more than 97% of publications, as is usual when consulting international scientific databases. However, there are also works published in Chinese, French, Russian, German, Dutch, Portuguese, Spanish, and Japanese.

#### *3.2. Distribution of Publications by Institution and Country*

Regarding the countries which have carried out the most research on this subject, the USA stands out, with more than 22% of the total number of published papers, followed by China, with more than 6%, and finally the United Kingdom, Germany, and Italy, with slightly more than 5%.

The affiliations of the works do not essentially respond to the same order as the countries. Thus, it was shown that the ten institutions that have published the most papers are, in order, Danmarks Tekniske Universitet, Universita degli Studi di Trento, National University of Singapore, Tallinn University of Technology, Cornell University, Instituto Politecnico do Porto, Delft University of Technology, UC Berkeley, University of Electronic Science and Technology of China, and Tsinghua University. It can be seen that two are from the USA and two are from China. Table 1 lists the main institutions and their main keywords used.

Figure 5 shows the relationship between research in this area in different countries using the VOSviewer software. In the network visualization of Figure 5, countries are represented by a circle. The size of the label and the circle of an item is determined by the weight of the country. The higher the weight of a country, the larger the circle and the label of the country. These clusters or communities are shown in Table 2. There are six communities of scientific collaboration where, in principle, the apparent lack of affinity between countries is striking, except in the case of community 3, which corresponds to all European countries. The cluster name is selected by the country that has the greatest weight within the cluster. Figure 5 shows the grea<sup>t</sup> centrality of the USA in this area, and although they belong to other clusters, China and the UK also occupy important positions of centrality. The clusters are led by Japan, Germany, Italy, UK, USA, and China, which, as can be seen, are the most industrialized countries in the world, all of them belonging to the G8 countries except China. For example, in the case of United Kingdom, the relationship is concentrated mainly with Canada, Iran, and Spain.


**Table 1.** Main institutions and their keywords.

**Figure 5.** Relationship between countries publishing on smart contracts.


**Table 2.** The detected clusters of countries related to research on smart contracts.

#### *3.3. Main Areas of Knowledge and Keyword Analysis*

The analysis of the keywords with which the works are indexed is one of the most relevant aspects in bibliometric analysis [38,39]. If the results obtained are analyzed according to keywords, and a cloud word is made with on-line software (see Figure 6) a strong relationship can be observed between blockchain technology, smart grids (SGs), and smart energy grids, as well as virtual currencies or electronic money (Ethereum, Electronic Money). It is worth highlighting the main areas of knowledge in which research is being carried out.

**Figure 6.** the cloud of keywords used in the work on smart contracts.

Figure 7 groups the keywords by large areas of knowledge, according to the Scopus indexation (subject area), and it can be seen that the first one, as was foreseeable, is computer science, followed by engineering via the theme of intelligent networks. The following areas of knowledge are those of energy and, later, of social sciences, among which the studies in law are framed. Indeed, if one looks at the keywords most commonly used in each country's publications (Table 3) one can see that blockchain and smart grid dominate in almost all the major countries with scientific publications on this subject.

There are also two keywords that, a priori, could go unnoticed, commerce and smart cards, which have a grea<sup>t</sup> relationship with the categories of social sciences and business, management, and accounting. This is particularly evident in the analysis of smart contracts in their focus on brokering, their transparent nature, the promise of greater commercial efficiency, lower legal and transaction costs, and, above all, the apparent advantage of anonymous transactions [40].

**Figure 7.** Thematic distribution of works related to smart contracts.


 Demand Response

 Computer Science

 Decision Making

> Commerce

 Electric Power Transmission Networks  Energy Resources

 Smart Power Grids

 Distributed Computer Systems

 Internet of Things

 Costs

**Table 3.** Countries and their three main keywords related to smart contract.

#### *3.4. Community Detection: Analysis of the Interconnection Between Keywords*

Power Grids

 Blockchain

 20 1.93

 18 1.74

 17 1.64

 16 1.55 Computation Theory

 16 1.55

Denmark

Switzerland

India

Iran

South Korea  Smart Power Grids

 Smart Power Grids

 Smart

Considering a community as a system composed of multiple interdependent elements, with a very wide range of relationships and intensities that are highly variable and dependent on each other, we could accept, conceptually, that communities are made up of a highly cohesive central core and peripheral spheres with unions increasingly weaker compared to the center. The central core would be structured by the most significant elements of the community, in terms of granting a definable individuality, representing the links between its constituents, and the strongest and most significant elements within the entire community complex. Communities or clusters are usually groups that are more likely to relate to each other than to members of other groups. When this analysis was completed by collaboration between authors from different countries using what is known as community detection, Figure 8 was obtained. An available online application, called VosViewer, which was developed specifically for this type of analysis of scientific production, was used for this purpose [41].

Clusters, or communities in networks, are one of the most notorious aspects of leading bibliometric studies [42]. These communities are groups that are more likely to be interconnected with each other than with members of other groups [43]. By analyzing the keywords of all the works published on smart contracts with the application calibrated for community detection, Figure 8 was obtained, in which six clearly differentiated communities have been detected (Table 4).

**Figure 8.** The relationship between keywords in the smart contract works.


**Table 4.** Detected clusters of keywords related to research on smart contracts.

The largest volume is cluster 1 (red), which groups 26.9% of keywords. The main keyword is smart cards, which has the highest density relationship with authentication and internet within its cluster, as well as cryptography (cluster 4) and contract (cluster 2). The first works related to smart cards from 1998 are mainly for use in electronic commerce [44].

Cluster 2 (green), which groups 23.6% of keywords, is focused on contracts. Its clusters highlight its relationship with laws and legislation and trade (marketing, purchasing, or project management). The relationship with other clusters is mainly through economics (cluster 5), cost or sales (3), smart cards (cluster 1), and blockchain (cluster 4).

Cluster 3 (blue) is focused on smart power grids. In general, one could say that this cluster is fairly independent, since its main relationships are within its own cluster. Thus, its main relationships are with electric power transmission networks, electric utilities, electricity market, commerce, electric load, energy management, and wind power. With other clusters, its main links are through blockchain and game theory (cluster 4), economics (cluster 3), and energy resources (cluster 6).

Cluster 4 (yellow) is mainly grouped around blockchain, which has the highest density relationship with electronic money and bitcoin within its cluster. Blockchain is focused on programming and cryptocurrency, such as Bitcoin. This is because cryptocurrencies and smart contracts are based on the same technology (blockchain) [8]. Regarding the relationship with other clusters, we find a connection mainly with commerce (cluster 3), economics (cluster 5), and security data (cluster 1).

Cluster 5 is grouped around economics, human, organization and management, or decision making. The importance of this cluster is that although it is not particularly important in terms of weight (8.5% of keywords), it is centered on the network, which shows that it is a link-point for all this research. This result possibly suggests that, in the near future, smart contract applications will play an important role in the different domains of modern organizations [45]. The other clusters highlight their relationship with blockchain (cluster 4), contracts (cluster 2), and smart power grids (cluster 3). Within its cluster, the main relationship is with terms based on the social economy, where key words such as human, male, female, adult, investment, transparency, and decision-making stand out.

Based on the previous study of clustering of keywords by clusters, its temporal evolution can be analyzed (Figure 9). It can be seen how clustering has temporarily evolved since smart cards and contracts to mainly two lines—first, to electronic commerce or quality of service, and second, to costs; from the first line (electronic commerce) to blockchain or network security and electronic money, and from the second line to cost energy issues, mainly regarding smart power grids. This gives an idea about the transition of the worldwide research in this topic.

There is no doubt that, despite the objections that can and must be made, bibliometric studies facilitate the understanding of research activity in a given scientific field. In this research, it has been observed from the analysis of keywords and the scientific communities that support them, that there is not ye<sup>t</sup> a community that dedicates itself to legislation and laws, despite being an essential aspect of the subject in which it is concerned. However, it should be noted that bibliometric analyses are generally valid in those areas in which scientific publications are an essential result of research. For this reason, the validity of bibliometric analyses is of maximum relevance to the study of basic areas, where scientific publications predominate, to a lesser extent in technological or applied areas, and to a much lesser extent in the areas of social and legislation. Therefore, comparisons between thematic areas and within these, scientific communities, should be made with caution, because the publication habits and productivity of authors differ according to knowledge areas. This is the case of this study, where differences were found between the areas of social sciences or business, managemen<sup>t</sup> and accounting, and those of computer science or engineering.

In the absence of a greater degree of maturity and extension of use, smart contracts raise several issues from a legal standpoint. Our law does not contemplate them, nor do judicial precedents ye<sup>t</sup> exist to help in this regard. However, it must be made clear that general contract law does provide criteria for verifying whether a smart contract can be legally valid and enforceable. The legal systems of our environment recognize the autonomy of the parties to freely reach legally enforceable agreements and contracts in the terms they consider, provided that the basic requirements of contract law are met, both in content (being a legal object and not a contravention of mandatory legal rules, ensuring the existence of valid consent of the parties, and obeying a legal cause) and in the manner of formalizing them.

**Figure 9.** The relationship between keywords in smart contract works and its evolution.
