The absence of applied research along with the diversity of definitions hampers the work of identifying common criteria for assessing digital skills and abilities. Finding a single definition with respect to the digital paradigm may well prove to be an arduous task. The impact is so broad and profound that the difficulty of assembling the whole puzzle of this new paradigm is quite impressive as each area and scientific discipline has adapted in accordance with its own rules and traditions.

A scientific revolution occurs, according to Kuhn (2008), when scientists find anomalies that cannot be explained by the universally accepted paradigm and the discipline becomes embroiled in a state of crisis. In the terminology used by Kuhn, a scientific revolution or a paradigm shift occurs precisely as a result of adapting to new realities that we wish to learn about.

A general but precise definition of digital paradigm can be found in IGI Global’s Dictionary Search:

Pattern of reality that results from a sampling process encoded in binary language. Unlike the analog paradigm, characteristic of a physical reality composed of complex signs that are not yet computable, the digital paradigm is reflected in a virtual reality composed entirely by computable finite groups of just two different signals (0 and 1).

It is quite difficult to separate the theoretical concept and the technological tools that gave rise to them on the basis of the definitions put forward in the literature. Heeks (2016) conceives of the problem just at the moment when this paradigm is starting to be widely adopted in the following terms:

...a digital development paradigm which conceptualizes ICT not as one tool among many that enables particular aspects of development, but as the platform that increasingly mediates development.

The same author details three phases of development for the new digital paradigm from a chronological perspective: 1) the pre-digital paradigm, 2) the ICT4D paradigm, and 3) the development of the digital paradigm (Figure 1).

Figure 1
Development and chronological shift of digital paradigm.
Adapted from Heeks (2016) original.

The pre-digital paradigm has lasted 50 years and has conceptualized the separation between new ICT and development (Heeks, 2009). The ICT4D paradigm began in the 1990s and promoted new conceptualized ICT as a useful tool for development. The paradigm emerged as a symbiosis between the onset of the generally availability of the Internet as a tool that is the basic condition for generating benefits and uses and the objectives for millennial development. And lastly, the digital paradigm, which took off at the beginning of this century, will continue over the years ahead and reach its plenitude at the end of the 21st century.

It is impossible to understand the new technological revolution or the consolidation of the digital paradigm without reviewing the evolution of the cellphone from its beginnings until the advent of 5G. First, we need to run through the history of its predecessors (1G, 2G, 3G, and 4G) to understand how our current reality was built up from these earlier generations.

Although new ICT go all the way back to the 1940s, it is really in the 1980s when 1G started to be used for mass wireless communication. This enabled us to make simple calls from one cellphone to another. The data transfer rate was then about 0.01 MB per second.

Ten years later (precisely in 1991), 2G technology was introduced, offering greater security through the use of digital encryption, unlike analog signals, and a faster rate of transfer of up to 3.1 MB per second, which was triple the speed offered by 1G technology. It also enabled regular text-only SMS messages to be sent.

Seven years were to pass before 3G technology was achieved and it was precisely in 1998 when the smartphone revolution kicked off, offering speeds of up to 14.4 MB per second and connecting mobile devices to Internet.

Almost a decade was to go by before the current standard, 4G / LTE, came into being in 2008. The most notable change was the incredible leap in terms of speed to 300 MB per second, which enabled the users to take part in activities like the transmission of high-definition audiovisual content.

In the last decade all effort regarding ICTs has focused on launching 5G. This offers the same basic features as its predecessors (text messaging, cellular voice calls and Internet connectivity), largely bolted on to the core 4G LTE Long Term Evolution) technology. Nevertheless, 5G technology stands out for four notable changes: bandwidth, which is expected to reach 1 GB per second; latency times - delay in the real process of data transfer – have been reduced to less than a millisecond; greater energy efficiency and an exponential increase for network connectivity.

The relevance of speed is self-explanatory, without the bandwidth, or size of the data packets transferred, being significant at any given moment. Increased energy efficiency is relevant since it significantly reduces costs and increases battery life, while also reducing CO2 emissions (Lope, Vidal & Mamaqi, 2018).

That said, the most important change is that of increasing network capacity since this forms the bedrock for the development of the Internet of Things (IoT). This all involves the creation of new infrastructures and a spectacular amount of growth in potential demand on the part of all types of users.

Table 1
1G vs 2G vs 3G vs 4G vs 5G: basic features.

Authors’ original work.

As can be easily discerned, the development of the new digital paradigm is marked by the 5G revolution and mobile technology.

Figure 2
Digital Paradigm and 5G.
Adapted from Heeks’ original work (2016).

Apart from being faster and more stable, and reducing latency from 1 to 3 milliseconds, 5G technology will enhance user experience (UX) by allowing mass use of virtual reality, augmented reality, IoT, tactile Internet (with applications in medicine, for example), automatic vehicles, and smart cities, among many other options. Of course, the question of whether we are prepared to face these drastic changes inevitably arises. Digital reality involves a permanent, ever-increasing link between users, companies, governments, etc. There should be full integration on the part of new digital professionals to this new reality.

In the digital business world, one of the greatest challenges raised by this new situation is that of making the most of the huge quantity of data - big data – to establish patterns for immediate reaction, the capacity to make predictions and enhance user experience. This all has a direct impact on the matter of digital competence.

In this respect, in 2006, the European Parliament and the EU Council (European Parliament and the Council, 2006) defined digital competence as follows:

Digital competence involves the confident and critical use of Information Society Technology (IST) for work, leisure and communication. It is underpinned by basic skills in ICT: the use of computers to retrieve, assess, store, produce, present and exchange information, and to communicate and participate in collaborative networks via the Internet.

In this definition the use of ICT has a prominent role in four large areas: for information, work, leisure and communication, subsequently highlighting the following skills: using computers, retrieving, reproducing, and saving information, skills for exchanging information and communication via Internet. The European Digital Competence Framework for Citizens sets out five macro-areas: i) information and data literacy, ii) communication and collaboration, iii) digital content creation, iv) safety, and v) problem-solving. Included in basic digital skills are technological skills, creativity, critical thinking and evaluation and effective collaboration (Ala-Mutka, 2011).

In practice, it is difficult for emerging digital professionals to find solvent scientific and institutional references for rigorous intellectual thinking and stringent ethical demands, a catalog of internationally recognized specific skills and abilities (Russom, 2011).

Among today’s various digital professionals – taking into account the most immediate needs for companies and their growth expectations – the five most requested profiles in recent years are Digital Marketing Manager, Digital Community Manager, Communication Manager, and Digital Analyst, as well as other emerging professional profiles that are configured on the fly in accordance with specific needs (Zbigniev, Chand Seal, Leon & Wiedenman, 2017).

The emerging discipline of Data Science covers the analysis, display and management of large sets of records (where ‘large’ in this context means many millions or billions of entries) The digitalization of all kinds of information, increase in the number of sensors and cheap storage have combined to generate huge amounts of data that are of interest to the Social Sciences sector and also to businesses (De Mauro, Greco, Grimaldi & Ritala, P., 2018).

Since the past decade, several authors have described the statistical work of data scientists as the sexiest job of the next 10 years (Granville, 2014). Forbes magazine describes the role of the data scientist as being the new concert in technology; and that Data Science is where geeks go (Marr, 2016). Many managers of large companies have already warned about the possible shifts that were going to take place but it is only recently in the middle of this decade when his premonition has proved to be correct (Miller, 2014). According to the EPYCE report database (2017) significant changes for professional technology families have come about since the biennium 2016-2017 and with forecasts for the next four years, until 2021. As can be seen in Table 2, many of the professionals that were linked to Technology in 2016 have disappeared or their name has been adapted to evolving reality, with the profession of Data Science Analyst emerging in 2017.

Table 2
Technology and adaptation of new professional positions.

EPYCE report (2017) and own elaboration.

In the immediate future, the professions in greatest demand and most difficult to cover, as can be seen in Graph 1, concern jobs involving technology and data processing.

Graph 1
Professional families with Jobs in greatest demand and hardest to cover.
Own elaboration (EPYCE database, 2017).

The positions of Big Data and Data Science are the most requested, according to forecasts for 2021, ranking first and second in a total of 32 positions of different professional families (in Graph 2 the top 12 positions are shown for forecasts for 2021). There are a total of 12 positions that make up almost 42% of the increase in demand in the labor market for 2021 whereas 20 positions only increase by 0.05% and some completely stagnate like the positions pertaining to Human Resources and Technical Marketing Engineer.

Graph 2
Forecast for increase in Jobs in greatest demand.
Own elaboration (EPYCE report database, 2017).

As can be observed, the most recent profiles have emerged from the discipline of Data Science. The Data Scientist is, in turn, the profession that best aligns with the digital paradigm.

In a search for jobs as a Big Data Analyst, only in one of the nation-wide technological employment portals is it revealed that in the past month the need for a total of 621 jobs related to these professions is announced (Figure 3).

Figure 3
Jobs for Big Data Analysts and Data Scientists announced in the past month.
Authors’ original work (October 2019).

However, once the need has been detected, its development and cataloging require rapid progress (Mamaqi, Miguel & Olave, 2011). Data scientists use specialized techniques to examine these troves of information to discover new ideas and create new value (White, 2012). In this profile, skills from different spheres and specialties of knowledge are grouped together, such as mathematics, statistics, economics, computer programming, business administration and advertising (Naur, 1966).

For the moment, these professional profiles do not require specific training at a Spanish university (most studies are for computer programmers or part-time or non-integral training), although unregulated courses can also be found. In practice, these professionals are trained on the fly and from the accumulation of shared experience. The best contributions in this regard are to be found intermingled with loose contributions from independent professionals, consultancies, etc., without a line of qualifications being set up to enable a complete teaching-learning process to be defined from start to finish and to ensure that it can be adapted to the digital jobs required in the market (Mamaqi, Marta-Lazo & Pérez, 2019). Data Science has, for many, proved to be a discipline of recent creation, with this role being assigned in the digital age in 2008 to Patil and Hammerbacher, data analysts from Facebook and LinkedIn respectively. Both predicted the importance of these professional profiles, immersed in huge amounts of data, organizing them, drawing conclusions so as to improve company operations and give them a key role in the decision-making process in various areas (Russom, 2011). However, its origin is much more ancient since Tukey (1962 and 1977, Naur, 1966) should be remembered when explaining the evolution of mathematical statistics. Thus, Data Analysis needs to be defined for the first time as a procedure for interpreting them. The aforementioned authors proposed ways to plan the collection of data to make analyzing them much easier, more precise or accurate and ways of suggesting hypotheses to be tested in statistical models. In 1996, the term Data Science was used for the first time at a conference entitled “Data Science, Classification and Related Methods”, organized by members of the International Federation of Classification Societies, where the statistical work was described as a trilogy consisting of re-collection of data, data analysis and modeling, and decision-making, requesting that statistics be re-named Data Science and statisticians Data Scientists. And at the beginning of the century, Data Science was introduced as an independent discipline, expanding the field of statistics to include the progress made in data computing. In 2012, at another international science event (The International Council for Science: Committee on Data for Science and Technology) CODATA (The Committee on Data for Science and Technology) was set up, redefining Data Science as Data Science and Technology.

De Mauro, Greco, Grimaldi & Ritala (2018) group the positions around Data Science professionals as: i) expert in data tools, ii) programmer, iii) quantitative statistician/analyst, iv) researcher v) data hacker vi) auditor vii) data protection and ethics manager and viii) data and strategy manager.

From what has been described above, at a minimum, it can be seen that these professionals more than comply with the basic digital skills and abilities in regulated education, with a certain level of perfectionism being achieved in Master’s and PhD degrees. Conceptually, the skills and abilities required of Data Science professionals can be divided into two groups:

1. 1. Technical and methodological skills and abilities related to the use of technological, statistical and analytical tools that are suitable for studying huge databases. This group requires multidisciplinary training that can mold the different skills and abilities associated with specific knowledge pertaining to data analysis.
2. 2. The capacity to enable value transformation using management techniques and knowledge of a specific business domain. This group includes soft skills, mainly in the area of communication.

With reference to the first group we have Cleveland (2001), who established six technical areas making up the field of Data Science: multidisciplinary research, models and methods for data, data computing, pedagogy, evaluation of tools and theory. Adapting these six areas to the skills of the first group within the professional Data Science family requires three positions: a Data Scientist, a specialist in Big Data and one in Data Analysis.

Based on these technical and practical contributions, in the following table we can see the list of requirements for top-level skills and specialties.

Table 3
Professional competences and capacities: Data scientist vs. Big data analyst vs. Data Analyst.

Own elaboration (July-October 2019 period).

As Bertalanffy, Ross Ashby, Weinberg et al. (1978) recalled in the 1970s, scientific and philosophical thinking has historically swung between systemic and holistic positions postulating, in the words of Aristotle, that the whole is greater than the parts (p.29) and that the Cartesian and positivist conceptions emerging in Europe in the 16th and 17th centuries gave rise to the industrial revolution. These latter conceptions are seen to become efficient once the problems have been broken down – fragmented, separated and distinguished – in such a way that complex phenomena can be analyzed by reducing them to their elementary parts and processes (p.31).

We are compelled to recognise the fact that, from now on, the processing of big data can serve both conceptions at the same time, and without implying any theoretical or methodological contradictions (Lope, 2018, p.146). Studies based on big data can establish correlations between phenomena that, in principle, are not thought to be connected, thereby illuminating new systemic visions. In turn, analysing such data will enable objective causal connections to be detected. Hence, thanks to technological progress, the two most significant trends of Western science are seen to be facing a theoretical reformulation with far-reaching consequences.

It is therefore imperative to recognize the new epistemological situation – new episteme derived from the latest version of téknē – and take advantage of it in all its potential for scientific research, particularly in the field of Social and Legal Sciences and Humanities.

That said, certain metaphysical inductions in the use of big data from a system or holistic standpoint should be avoided. As Google’s ex-data scientist Stephens-Davidowitz (2019) usefully points out, the temptation to make predictions after detecting simple correlations of events usually leads to failure as it should be borne in mind that many correlations are nothing more than fortuitous coincidences and do not prove any causality whatsoever. The solution is not always a case of finding more big data. A special ingredient is often needed for it to work best: human judgment and small-scale questionnaires, i.e. something that we could call small data (Stephens-Davidowitz, 2019, p. 253).

This theoretical innovation we are trying to outline cannot ignore the fact that the hugely expanding incidence of innovations in people’s everyday life has ethical implications. We are apparently offered things and satisfactions when, in practice, such things and satisfactions require us to adopt a different lifestyle. The imposition is received through our affective channels, as Mark Hunyadi (2015) cleverly points out. In this way it appeals to us, or at least something about it appeals strongly enough for us to accept the rest of it that is less appealing. According to this model, a system gradually manages to eventually impose a lifestyle that no one has explicitly wished for, that perhaps nobody would have wanted if they could have chosen outright (p.70).

It is essential to place something in the theoretical middle that we tend not to want to see. This is because the technique is never neutral from an ethical standpoint. It is not true that a tool’s goodness or evil depends solely on the use – whether beneficial or destructive – that is made of it. (Mamaqi, Marta-Lazo & Pérez, 2019).

Judy Wajcman (2017), the specialist in sociological studies on the impacts of technology on people’s lives does not mince her words, saying that the time has come to question the euphoria for speed and the technological impulse to achieve it, by harnessing our inventiveness to gain control of our time for a bit longer (p. 257).

Hence, the need to assume, both regarding theory and methodology, the initial need to carry out a thorough analysis, just like any scientific task, applied to all types of tools and services, irrespective of whether they are real or virtual. Algorithms, robots, and applications should be subjected by the scientific community to rigorous studies that evaluate their characteristics taking into account that we are dealing with complex texts we engage with and which participate in the configuration of our subjectivity in registering what is real, what is imaginary and what is semi-cognitive and which, occasionally can even have a symbolic dimension (Lope, 2018).

The theorization of communication processes itself requires the rethinking of such tenets since, by and large, the Shannon and Weaver Model of Communication now only applies to the problems for which it was originally conceived, i.e. for transferring information from one machine to another. On the other hand, communication processes among humans – whether employing secondary, tertiary, or quaternary means – are far more complex than the theoretical simplification proposed by Shannon and Weaver.

It is no longer acceptable to entertain any model of human communication that does not incorporate the recent discoveries made in neuroscience. Among such discoveries is the one detected by Freud and confirmed by current research that the origin of our decisions and functions – communication being one of them – are unconscious. From this we can derive a very important notion: the idea that a subject already exists prior to the communicative act is no longer defensible in any circumstances. If unconscious desires do exist, this idea that there is a subject that is fully aware when taking a decision to issue a message is fallacious. The Internet of Things does just that: it amasses mountains of data on the subjects’ unconscious activities.

In fact, the most important thing about human communicative processes, namely meaning, is the most difficult to pin down in theoretical terms. Gonzalo Abril (2005) reminds us that meaning is immeasurable and paradoxical. Since, unlike what happens with the information processed by IT, it is not possible to generate a metalinguistic discourse for meaning, because no external discourse exists outside meaning itself (p.36). And for the real experience of each subject with signals, images, speech, and objects, the really decisive thing is meaning. As explains neuroscientist Antonio Damasio (2018), the merely cognitive level must be distinguished from the emotional level even though both things are practically inseparable in the individual’s subjective experience. And meaning, according to Damasio, should be understood as the real value «felt» by the human being from life’s consequences, good or bad (p.31). The alliance between Data Science and Neuroscience should provide ways to integrate meaning and subjectivity in science.

Back in the early 1970s the philosopher and sociologist, Julien Freund, explained the dangers involved in ideological deviations that were undermining the scientific nature and rigor of the Social Sciences and Humanities. In the first two decades of the 21st century, there is no sign of improvement but rather of uncritical accommodation within formulaic expressions – largely generated as a result of political correctness – which are offered as valid concepts without even devoting any intellectual effort to demonstrating their theoretical validity (1975). What Freund (1975) was saying now takes on undisputable value and relevance: The human sciences specialist does not only work with objective facts but also with opinions, beliefs, meanings. Their object consists of human actions and they cannot forget that people give them significance, even when they are unwanted or unprepared (p.147).

Now it is possible to design analytical methods capable of producing cartographies that are fairly objective regarding the reality of, among others, social, communicative, economic, emotional, artistic, and political processes. Those analytical methodologies based on big data should be guided by scientific and ethical rigor in such a way that, as they go about enhancing our knowledge, they also provide protection for the subjects themselves that are generating such data without, in most cases, even being aware that they are doing so.

If with regard to university training programs it is essential to make a huge effort to achieve multidisciplinary coordination and creativity in order to offer the Bachelor’s and Master’s degrees required, there is another task that is even more urgent: that of guaranteeing the highest quality and innovation in scientific publications in general, and in the fields of Social Sciences and Humanities in particular. And this should be done by unequivocally adopting the advantages that are already to be had from big data and AI. The inalienable objective that Data Science makes available to publications and researchers in general is to incorporate methods for evaluating the effective content of what they publish for quality and, in the process, detect whether what is published offers real innovation. By and large, current methods that assess the quality of journals and editorials are based on formal qualitative and quantitative aspects, which have been used to rate the scientific production of the country, research bodies and even the researchers themselves. To date, the usefulness of such procedures and their ratings has not brought about agreement among the specialists on evaluating scientific documents nor does it satisfy many authors (Aguillo, 2015). Furthermore, the scant attention paid to publications on Social Sciences and Humanities has led to the creation of new tools and measures that also pose problems, since all researchers are aware that the privileged position of a specific journal in some rankings will in no way guarantee the quality of the real intellectual interest of the content published (López, 2016). This state of affairs is a huge obstacle that seriously hinders scientific progress.

With respect to measuring the impact of the publications on the scientific community internationally, the Journal Citation Report, known to all and currently published by Clarivate Analytics, or the Scopus database, which through the facilitation of abstracts and Website performance indicators, endeavors to improve and complete the results of the WOS (Web of Science).

Bibliometric works are based on the use and analysis of data from scientific publications and have given rise to interesting fields of research such as Webmetrics and Alternative Metrics, which are beginning to yield hopeful results, complementing those mentioned above.

The development of Internet and the Web, digitalization and open access, are generating a huge amount of data, whose processing and analysis using AI techniques facilitate more reliable information on the impact of scientific publications as well as generating new knowledge.

With regard to qualitative procedures, peer review often exposes shortcomings like the lack of clear criteria for the appraisal of proposals or the generous deadlines it demands.

The use of specific algorithms base don Cartesian genetic programming enables the review process to be reduced by about 30% (Mrowinski, Fronczak, Ausloos & Nedic, 2017).

One fundamental element that affects the quality of periodicals concerns management procedures. The emergence of platforms for publishers of scientific works in open access, such as Frontiers¹, is an example because it focuses both on providing assistance to editors, reviewers and authors, as well as the technical development of tools that make improving the quality of the scientific publications easier at all levels, including the identification of potential reviewers. To do so, it uses AIRA (Artificial Intelligence Review Assistant), which harnesses AI and machine-learning techniques and helps editors and reviewers, by facilitating their interaction and offering rapid results. This safeguards quality, rigor and independence, which presumably provides a guarantee for the authors. AIRA is an instrument that is designed to assess the quality of the work and facilitate the identification of specialists to carry out the review. Its algorithms identify overlaps with other studies, plagiarism, and flag up the quality of language and other aspects. It is regulated by processes that need to be followed by those texts that satisfy the quality criteria, and others for those that do not meet such criteria. This all means there is a significant reduction in review times. Finally, the platform is generating new metrics that will make it possible to assess the impact of academic articles objectively, along with other resources such as Loop, a network of scientific researchers that facilitates the dissemination of published books and articles.

In the same vein, ScholarOne is a peer review platform, employed by the press, which uses UNSILO, a program that analyzes written manuscripts through natural language processing and machine learning. This allows principle statements and other functions to be extracted, thereby reducing the text to a script. It also detects similar ideas in other works, allowing any instances of plagiarism to be located and facilitating the localization of texts on similar topics. The application Wizdom.ai, which facilitates collaborative writing, is compatible with Google Docs, and uses databases, enabling connections to be established between disciplines and concepts, topic reviews, etc. Others like Penelope.ai examine the structure and citations of a text, as well as the statistical studies it contains. With this last function, StatReviewer is also used. These are tools that represent a huge step forward in offering contributions of much higher quality and also greatly reducing the amount of work required at the review stage. Since, in the latter case, their algorithm is capable of conducting a numerical assessment of a work.

Better known is the case of Open Journal System, which is starting to be used at universities and research centers both nationally and internationally, facilitating the various different management processes for such publications.

AI is also useful for performing repetitive tasks in journal editing, such as providing authoring assistance, storage, description and indexation in repositories, review by personnel not working for the journal, and dissemination on social networks, which can be coupled and decoupled in a systematic way (Priem & Hemminger, 2012).

Techniques like Natural Language Processing (NLP) and Machine Learning (ML) have not stopped developing and continue to do so today, being applied to all types of text, including scientific ones. The extraction of descriptive data and other metadata can facilitate rapid consultation and provide the most significant ideas.

In addition, the implementation of AI-based software, making it possible to describe both sound and visual documents by scanning them, can be of fundamental assistance in optimizing retrieval and dissemination of these types of documents, which very often accompany further documentation.

The new formats of scientific studies published online allow for the rapid, mass processing of data and its incorporation in author profiles and publication components. The incorporation and display of the data is immediate. Journal-publishing giants such as Elsevier are implementing such technologies, which also need to become better known and adopted by more modest publishers so as to continue with their dissemination work.

Such technologies enable a huge amount of usage data to be extracted from both documents and logs, which has created a new analytical horizon for these publications (Jamali, Nicholas & Huntington, 2005).