Introduction

Digital technologies are widely recognized as one of the most impactful and disruptive developments since the industrial revolution, thus profoundly reshaping the world we live in. As humankind enters the new era of the digital technology revolution, Digital Entrepreneurship (DE) is actively reshaping several industries and markets, while providing enterprises with new and unprecedented opportunities, thus prompting a massive surge in interest from both academics and practitioners. Theoretically speaking, DE refers to establishing a company through electronic platforms, thus offering products and services based on electronic value. In other words, information technology makes the business value created through DE possible (Kollmann, 2006). While several definitions of DE exist, it is generally accepted that digital technologies have breathed new life into entrepreneurship literature through disruptive innovations and the advent of new business models (Huang et al., 2017), thus generating a surge in interest from both academia and industry (Jawad et al., 2021; Park et al., 2021). In fact, digital tools enable entrepreneurs to enhance their company's performance further, regardless of the phase they are in, thus opening new avenues and opportunities never experienced before. As the boundaries of physical and digital businesses are seemingly becoming nonexistent thanks to the disruptive push of DE, it has become imperative for both academia and industry to investigate the phenomenon in depth and learn how to master it.

Much like adjacent fields tied to innovation and digital transformation, the field of DE has seen a significant rise in popularity throughout the past few decades. Thus far, scholars have investigated the numerous implications of digital technology regarding entrepreneurship. Their attention toward DE has spiked in recent years (Kraus et al., 2018; Zaheer et al., 2019), primarily due to the increasing relevance of digital transformation in today's landscape. From a theoretical perspective, the field of DE is positioned at the intersection of digital technology research and traditional entrepreneurship literature, as it focuses on entrepreneurial innovation processes and outcomes (Nambisan, 2017; Nambisan et al., 2019). The underlying assumption of DE is that the disruptive change brought about by digital technologies is unprecedented and incomparable to traditional technologies due to their generative capacity, disruptive potential, and constant evolution (Berger et al., 2021).

The literature on DE is fragmented, diverse and employs various choices in research design. As such, several authors have attempted to perform systematic reviews of the literature on specific DE topics (Kraus et al., 2018; Satalkina & Steiner, 2020; Zaheer et al., 2019), to provide a more jointed overview of the field. However, as Paul et al. (2023) pointed out, a bibliometric approach is currently needed to investigate the evolution of DE literature over the years, as it would provide insights currently lacking from DE research. In other words, what is currently missing is a comprehensive bibliometric investigation of the field, highlighting key bibliometric parameters, such as the most relevant scholars, most prolific countries of research, and the evolution of core topics over the years (Paul et al., 2023). Finally, aside from its uniqueness and originality, the significant volume of scientific production on DE makes adopting a systematic review unsustainable due to the impossibility of manually reading each document (Denyer & Tranfield, 2009).

Consistently with the gap identified above, we apply latent Dirichlet allocation (LDA) and topic modeling (Blei, 2012; Blei et al., 2003) to a sample of 201 documents. Topic modeling is popular amid entrepreneurship research, as it allows for precise and transparent processing of vast amounts of data. Furthermore, unlike other textual mining techniques, LDA does not require prior knowledge of the subject, thus enabling an unsupervised approach to the sample being investigated. Consequently, we identify several core topics through LDA in the sample, subsequently conducting a descriptive analysis of the dataset based on bibliometric information (Kaushik et al., 2023; Ligorio et al., 2022; Wang et al., 2022). The result is a deep understanding of how the DE literature has evolved over the years, thus defining its past, present, and possible future developments.

Our study provides several contributions to both theory and practice. From a theoretical perspective, it answers the call for a bibliometric review of DE literature made in previous studies (Paul et al., 2023). In doing so, our study is the first attempt to comprehensively structure scientific knowledge on the topic of DE through the use of LDA topic modeling, thus complementing existing reviews in the field (Yu et al., 2022; Contreras Cruz et al., 2022; Gil-Gomez et al., 2021), and providing unique insights through the use of an automated, machine learning-based analysis. Recently, Zhai et al. (2022) performed a bibliometric review on the topic of DE, yet while their contribution is of absolute importance for the field, they suggest that using different approaches to bibliometric analyses, in our case a big data, topic modeling-based approach, may yield different results, and provide new insights. Additionally, their sample is limited to studies published up to 8 March 2021 (Zhai et al., 2022), while ours provides a more updated look, especially given how prolific the field has been throughout 2021, 2022 and early 2023. In fact, roughly half of the studies featured in our sample were published after March 2021. Further, Majdouline et al. (2022) used a bibliometric approach to map technological entrepreneurship, focusing on deploying technologies with tangible applications. Similarly, Fernandes et al. (2022) mapped the field of digital platforms, yet their scoped approach did not capture DE as a whole. While previous systematic reviews of the DE literature exist, its ever-increasing scale and spread call for a bibliometric approach to capture recent and past developments more comprehensively.

Additionally, our attempt is the first application of topic modeling to structuring DE literature. Given the multidisciplinary nature of DE, we deemed topic modeling the most appropriate solution to achieve the most reliable understanding of the whole body of research. Finally, we strive to provide scholars with a clear picture of the present landscape of DE research, thus helping them better position their work to journals most interested in their articles. From a practical perspective, our study is of interest to entrepreneurs willing to explore the opportunities presented by digital transformation. Our bibliometric analysis intends to achieve this by presenting how DE has paved the way for novel business models to emerge, how it has fostered entrepreneurial activity, and how it has helped develop existing businesses.

Given the aim set above, we draw on established bibliometric research to define the study's research questions (RQs). A bibliometric study's primary objective is to retrospect the performance and science of a research field (Donthu et al., 2021), in our case DE. We define RQ1 accordingly, as we set ourselves the goal to map extant knowledge on DE from a bibliometric perspective, thus providing a comprehensive look at the field through varied bibliometric information, namely annual scientific production, citation distribution per year, most cited journals, and articles, and most prolific countries. Further, drawing on the work of Aziz et al. (2021), we extract the main topics through LDA topic modeling and illustrate them from a qualitative perspective to show how the topic focus has evolved over the decades, thus defining RQ2. Finally, we draw on the work of Nájera-Sánchez et al. (2022) to formulate RQ3 by adopting a critical approach meant to highlight research gaps and future directions in the field of DE. With the above in mind, our paper aims to answer the following research questions:

  • RQ1: What is the current scientific landscape around DE?

  • RQ2: Which themes involving DE are the most popular and emerging among scholars?

  • RQ3: What future research themes on DE can benefit from further academic attention?

The remainder of this paper is structured as follows: the following section provides a theoretical overview of DE, thus setting the conceptual boundaries of our research. Section “Methodology” features the methodology used in the study, clearly depicting the steps taken in the research protocol. Subsequently, Section “Results: The Current State-of-the-art of DE” discusses the findings in terms of bibliometric information and emerging topics. Later sections feature an overview of the research agenda developed based on extant gaps in the literature and a theoretical framework synthesized from actual knowledge of DE. Finally, the manuscript concludes with the implications, limitations, and recommendations for future bibliometric research.

Theoretical background and conceptual boundaries of the study

As a research domain, entrepreneurship is vast and diverse; as such it has been investigated by scholars belonging to different disciplines and cultures. In more recent times, scholars have attempted to delineate entrepreneurship as a distinctive domain of its own, thus contributing to the emergence of entrepreneurship as a field. Given the broad scope of entrepreneurship research, prior to performing the bibliometric analysis, it is imperative for us to define the conceptual boundaries of the study by exploring the definitions of the terms we will be employing, namely DE and its variants (Giones & Brem, 2017). The concept of entrepreneurial activity dates back to the early 1840s and the emergence of industrialization, albeit the first definitions of the term 'Entrepreneurship' became more widespread during the 1980s (Stevenson, 1983). According to Gartner (1989), an entrepreneur is the founder of a business organization meant to address the specific needs of a chosen market niche in exchange for wealth and independence. Overall, entrepreneurship is considered a broad definition, covering both small and large organizations, as business owners in any context and dimension see entrepreneurial mindsets as a means to achieve business development and innovation (Stevenson, 1983). Ratten (2023) attempted to review the plethora of definitions of entrepreneurship found amid research, with the intention of making sense of the conceptual evolution of the field. According to their review, the definition of Shane and Venkataraman (2000) is one of the most well known, and reads as follows: “how, by whom and with what effects opportunities to create future goods and services are discovered, evaluated and exploited.” While their definition puts great emphasis on the opportunities presented to entrepreneurs willing to make the most of them, more recent definitions have shifted attention towards the process itself of identifying potential business opportunities (Nambisan, 2017), which allows for a more flexible approach to entrepreneurship as a concept, and help understand how innovation and creativity have contributed to the evolution of the field over time.

Having reviewed the main definitions of entrepreneurship, we now focus on DE’s definitions specifically. In recent times, digital transformation has radically changed the global business landscape, bringing disruptive developments to entrepreneurship and virtually every other aspect of the business. Despite its popularity, the conceptualization of DE remains quite elusive. One of the very first attempts at conceptualizing DE was made by the European Commission (2015, p.1), posing that DE “embraces all new ventures and the transformation of existing businesses that drive economic and/or social value by creating and using novel digital technologies. Digital enterprises are characterized by a high intensity of utilization of novel digital technologies (particularly social, big data, mobile and cloud solutions) to improve business operations, invent new business models, sharpen business intelligence, and engage with customers and stakeholders. They create the jobs and growth opportunities of the future.”

Theoretically, it is somewhat unclear whether DE is considered a subset of entrepreneurship associated with technological innovation or a subset of innovation literature associated with entrepreneurship (Fischer & Rebecca Reuber, 2014; Sigfusson & Chetty, 2013), due to its hybrid and somewhat multidisciplinary nature. Currently, scholars use DE as a term to describe the category of entrepreneurship in which “some or all of what would be physical in a traditional organization has been digitized” (Hull et al., 2007). As of today, DE enables entrepreneurs to respond to an ever-changing technological environment effectively, thus strategically answering the needs and wants of their niche market as they change over time (Autio et al., 2018).

However, several more definitions of DE exist, as the field is wider than the ones mentioned above. For instance, Shane and Venkataraman (2000) define digital start-ups as nascent organizations in which digital technologies are vital for their business model and not merely accessories. For instance, digital technologies might contribute to the products or services being offered, the value delivery processes, or the revenue streams, thus making them essential for the company’s life. A further definition of DE is formulated by McMullen and Dimov (2013), who conceptualized DE as the process of creating a digital start-up as a new business or within a company that has already been established. Hull et al. (2007) defined DE as “… a subcategory of entrepreneurship in which some or all of what would be physical in a traditional organization has been digitized”, while Sussan and Acs (2017) defined it as “… any agent that is engaged in any sort of venture be it commercial, social, government, or corporate that uses digital technologies.” More recently, Kraus et al. (2018) expanded upon previous studies, concluding that DE is still growing and in an early development phase, despite many authors attempting to conceptualize its definition (Davidson & Vaast, 2010; Sussan & Acs, 2017).

Further, the conceptual scope of DE is broader than the business strategy. Instead, it encompasses several aspects of entrepreneurial activity, including digital leadership, knowledge management, and digital creativity (Anwar & Daniel, 2016; Bach et al., 2018). Consequently, from a theoretical perspective, modern DE research strives to highlight the role played by digital technologies in entrepreneurship. However, limited efforts have been made to conjecture on the role of specific aspects of DE (Engel, 2015; Kollmann et al., 2022). Said conceptualization would be rather significant given its multifaceted nature and how intensively technology is reshaping entrepreneurship worldwide (Prashantham et al., 2018; Ojala et al., 2018), thus generating new terms, definitions, and meanings. For instance, the definition of the “Digital Ecosystem” is somewhat new (Khlystova et al., 2022) and is currently extensively used in academic research. A further example of a conceptual field adjacent to DE is that relating to digital platforms, namely infrastructures or artifacts operating through the power of digital computing (Nambisan, 2017). Modern examples of digital platforms are smartphone apps, Internet of Things (IoT) solutions, and websites. Regarding DE, digital platforms act as enablers, as they allow digital entrepreneurs to use them to perform their entrepreneurial activities.

Methodology

Our research design is based on the bibliometric analysis of the DE literature to understand the evolution of scientific production on the topic over the decades. Drawing on the protocol set by Donthu et al. (2021), we consider bibliometric research to be an established methodology in entrepreneurship research, as authors have used it to summarize scientific production on a topic of their choice comprehensively. Prior research has applied a bibliometric approach to entrepreneurship (Yu et al., 2022; Contreras Cruz et al., 2022; Gil-Gomez et al., 2021), yet an attempt has yet to be made thus far to explore DE specifically. Consequently, after carefully considering the manuscripts mentioned above, we adopted a research protocol starting from the definition of the conceptual boundaries of the study and followed by several sequential steps depicted below.

Drawing on previously published bibliometric works, we performed several analyses to map scientific knowledge on the topic. The analyses were divided into two distinct sections, one quantitative and the other qualitative. The quantitative approach illustrates the field from a macro perspective, highlighting bibliometric data, including annual scientific production, citation distribution per year, most cited journals and articles, and most prolific countries. Consistent with our goal of presenting the state of the DE field, we performed two distinct science mapping analyses for clear data visualization, namely keyword co-occurrence. Following previous studies, we used VOSViewer to visualize data and conduct the analysis.

Data collection

The Core Collection of Web of Science (WoS) provided us with the bibliometric information used in the analysis. The decision to use WoS specifically, thus excluding other databases, was drawn from previous bibliometric reviews published on entrepreneurship. Contrary to other directories, WoS is “the longest-running citation index” (Finch, 2012, p. 246). Thus, it is the optimal choice when attempting to cover the entire development of a specific literature stream. Furthermore, WoS is the most used repository for bibliometric analyses, as it features a comprehensive range of prestigious social science and entrepreneurship publications. As for the decision to exclude more databases from our search, according to previous studies, it does not equate to more precise results (Aziz et al., 2021; Harzing & Alakangas, 2016). In other words, WoS provides a representative enough sample to conduct an accurate analysis of the evolution of DE without the need to compare the dataset with information found in other repositories.

Drawing on Zupic and Čater (2014) and having established the reasons why we opted to focus on one database specifically (Aziz et al., 2021; Harzing & Alakangas, 2016), we compared our database of choice, namely WoS, with alternative repositories. Despite its significant scope, Google Scholar lacks a user interface or application programming interface (API) that would allow us to extract bibliometric data, thus making it unsuitable for bibliometric analysis (Zupic & Čater, 2014). Given the wide range of records, Scopus is deemed a valid alternative, applicable when mapping smaller research areas that WoS would cover insufficiently (Li et al., 2018). However, in the case of DE, we deemed the number of records extracted from WoS sufficient in terms of data saturation (Zupic & Čater, 2014), as they would fall within the range set by previous bibliometric research in the field of entrepreneurship (Aparisi-Torrijo & Ribes-Giner, 2022; Deyanova et al., 2022; Martínez-Climent et al., 2018). Consequently, we deemed our choice of WoS, the leading database in terms of scientific citations and analytical information (Li et al., 2018), as the most appropriate for the research.

We conducted an initial search with the term "digita* entrepreneur*", in titles, abstracts, and keywords (Fernandes et al., 2022). Using a search term in titles and abstracts allows for filtering out studies not related to the conceptual scope of a bibliometric study (Kalantari et al., 2017; Kraus et al., 2018). Additionally, by searching for the term in the "Keywords Plus" category, we could add further relevant entries through the WoS algorithm. An initial search with the above search string resulted in 405 publications, ranging from December 1993 to December 2022. Given the broad scope of DE definitions (Zhai et al., 2022), narrowing the search to “digita* entrepreneur*” raised some challenges. While the primary search was conducted through the use of “digital” and “entrepreneurship” keywords, we integrated the search with a supplementary search formula featuring “digit* start up”, “online startup”, “e-startup” “digit* startup”, “cyber entrepreneur*” and “internet entrepreneur*”.

The dataset was consequently narrowed by focusing on the categories "business", "management" and "economics", thus filtering out articles not related to the social science field. Additionally, we excluded articles not written in English, for consistency across the sample. Similarly, we excluded non-peer-reviewed sources, namely book chapters and conference proceedings, to ensure academic rigor. In other words, we limited our sample to the WoS categories of "article" and "review", as established by previous works (Merigó et al., 2016; Merigò & Yang, 2017). Consequently, our sample was narrowed down to 201 entries. A careful examination of the sample revealed no duplicated records or papers that, despite fitting the above criteria, were outside the conceptual boundaries set for the study. We have, therefore, finalized the sample of 201 documents to be investigated, which is in line with previous bibliometric studies conducted in the field of entrepreneurship (Aparisi-Torrijo & Ribes-Giner, 2022; Deyanova et al., 2022; Martínez-Climent et al., 2018) and with the cut-off value set by Rogers et al. (2020) for bibliometric studies. Drawing on Cano-Marin et al. (2023), Fig. 1 summarizes the steps mentioned above.

Fig. 1
figure 1

Flowchart describing the sample selection process

Full size image

LDA topic modeling

Topic modeling enables researchers to unveil latent topics amid a textual corpus of documents. In other words, topic modeling allows researchers to synthesize significantly large datasets in a selected number of distinct clusters. Drawing on previous scientific works (Kaushik et al., 2023; Ligorio et al., 2022; Wang et al., 2022), LDA topic modeling allows researchers to process large quantities of documents in an objective and replicable way, thus making it the most appropriate choice for our bibliometric research design. In order to perform the analysis, researchers need to determine both the number of topics to be extracted and the content of the topics themselves. The first step is achieved through probabilistic modeling, along with a qualitative review and labeling of the extracted clusters. The second step is performed through the extraction of several models, each with a different set of topics, followed by a manual comparison of the output.

While several topic modeling techniques exist, LDA is the most widely used amid social science research. More specifically, scholars consider LDA a valuable tool for bibliometric research due to its design, as it allows researchers to identify a mixture of topics and terms, with terms being shared across multiple topics. This is ideal when it comes to the corpora of documents, as topics within specific subject areas may overlap, at least partially (Blei, 2012). To understand LDA topic modeling, we can refer to Fig. 2.

Fig. 2
figure 2

Visualization of the LDA algorithm

Full size image

LDA uses word co-occurrence to determine a set number of topics amid a corpus of documents. In the case of our bibliometric study, an article might contribute to one or more topics, each with a distinct percentage of shared words. Each of the said topics features a list of closely related words left to the scholars to read, interpret and label. More precisely and for transparency and replicability, we will depict each step taken in detail through the paragraphs below.

Once we had collected and extracted bibliographic information from WoS, we used the R package LDAShiny to develop our LDA model. We first removed punctuation marks, special characters, and numbers, along with stop words from the sample. Stop words are common usage words that do not add context or value to the analysis. Hence, it is best to remove them for the sake of accuracy. We then created a document-term matrix (DTM), counting term occurrences per document, and performed term frequency–inverse document frequency (TF-IDF) analysis to remove low-frequency and high-frequency terms from the sample.

Consequently, we have run goodness-of-fit tests to determine the ideal number of topics to extract from the sample. While several goodness-of-fit tests exist, we focused on harmonic mean and coherence scores. By triangulating the analyses, we determined the most consistent number of topics to extract.

Harmonic means measure the intensity with which a certain distribution varies across topics. Higher values for harmonic means indicate an even distribution, while lower values suggest the distribution is unequal, thus concentrated in only a few topics among the sample.

Additionally, we calculated the coherence scores of models featuring different numbers of topics, ranging from 2 to 50. Coherence indicates the semantic similarity between the most recurring words within a certain topic. In other words, models with low coherence scores tend to share the same words across multiple topics, thus making them unclear and unreliable. Meanwhile, models with high coherence scores suggest that each topic is distinguished from the rest.

Running the LDA topic modeling tests

In order to perform the LDA tests, we made use of the R package LDAShiny. The first step was to determine the ideal number of topics to extract from the model. By comparing the results of our analyses, namely coherence scores and harmonic means, we identified n = 17 as the most appropriate number of topics to extract. As we can see from Fig. 3, models featuring more than 17 topics have decreasing mean scores. Similarly, when looking at Fig. 4, models featuring more than 17 have decreasing increments in coherence. We thus extracted the 17 topics from the sample and proceeded to their interpretation. The labeling of the topics is subjective and left for the researchers. We triangulated the results of our LDA analysis with the conceptual boundaries set by the study to understand them and provide an appropriate label for each topic.

Fig. 3
figure 3

Harmonic mean score

Full size image
Fig. 4
figure 4

Coherence score

Full size image

Prior to extracting the topics, however, a necessary step was to process the corpus of documents. In other words, we had to perform data cleaning on the bibliometric information extracted from the online repository in order to more accurately extract the information needed through LDA. We first began by tokenizing the sample and by removing stop words that offer no context to a specific topic (for instance, words such as ‘and’, ‘of’, ‘the’). We then stemmed the words remaining from the previous steps, in order to reduce the number of unique words and, once again, improve the overall quality of the model. In other words, we removed suffixes to make sure each word variant would be reduced to the common stem. The logic behind LDA topic modeling was first introduced by Blei (2012) and has proven effective for reviewing academic literature. LDA is ideal when attempting to synthesize large bodies of knowledge, thus being an ideal choice for a bibliometric study which employs a much larger sample size than traditional systematic literature reviews. It does have a few limitations, however, as it fails to go in-depth amid a smaller set of data, so it is not recommended when the sample size of papers is limited. This is true in the case of a systematic review. Table 1 features the 17 topics extracted from the sample, each with their own label that was manually reviewed and assigned by the authors. The description tab is formulated by the authors, as they interpret the most recurring words featured in the topic and provide a comprehensive depiction of its content.

Table 1 Topics Extracted through LDA analysis
Full size table

When looking at the 17 topics extracted through LDA topic modeling, we were able to identify four distinct clusters of research, namely: business model transformation, the implications of DE for innovation, DE as enabler of empowerment, and digital entrepreneurial platforms. Consequently, we used these clusters to structure the rest of our analysis and divide our observations into four distinct thematic areas.

Results: the current state-of-the-art of DE

Publication trends

When formulating RQ1, we asked ourselves what publication trends could be found amid the DE literature (Donthu et al., 2021). We analyzed several bibliometric indicators to answer this question, looking at scientific publications by year, country, source, and research design. Albeit that they are descriptive, reviewing publication trends in DE literature helps draw a comprehensive overview of the field and understand its evolution from a bird’s eye view.

Regarding annual scientific production, the earliest publication found in the sample dates to 1998. Ever since, the field has been growing steadily in terms of scientific output, with a significant increase from 2019 onward. The overall trend is in line with fields interlinked with digital transformation, as the proliferation of digital solutions available for entrepreneurs has inevitably contributed to the increase in attention from academics and practitioners. A possible explanation comes from the surge in interest towards DE brought about by times of uncertainty and pandemic crises (Elia et al., 2020, 2021), during which several entrepreneurial ventures were forced to consider digital technologies as enablers for their entrepreneurial ventures. Figure 5 illustrates annual scientific production in the field of DE, highlighting the cumulative growth of publications since 1998.

Fig. 5
figure 5

Annual scientific production on DE (1998–2022)

Full size image

Scientific publications on DE are scattered across the globe. Table 2 features the DE literature stream's geographic scope by depicting the most prolific countries in terms of scientific output, and the countries cited the most, thus those which are the most influential in the literature. The United States (US) leads the table for the most citations, whereas the United Kingdom (UK) appears to be the most prolific country in terms of published papers. Our interpretation of the data is that western countries heavily dominate the DE literature, as a significant majority of contributions come from there, thus possibly skewing the results towards the western conceptualization of entrepreneurship and leaving out other cultural perspectives. Consequently, future research could seek to develop empirical evidence based on eastern Europe and Pacific countries to complement the already extensive catalog of literature on DE. Similarly, cross-cultural studies could help mitigate the disproportionate sample by unveiling a new perspective on DE.

Table 2 Most active countries on DE by citation and number of documents
Full size table

More specifically, when looking at the peculiarities of each of the most prolific country’s scientific output, the UK presents a varied sample of contributions, ranging from innovation-focused papers (Nambisan et al., 2019), and studies on women entrepreneurs (Dy et al., 2016; Martinez Dy et al., 2018). Surely, the latter is one of the most prominent patterns of research that can be derived from UK-based scientific production, as several of their contributions explore the role DE plays in emancipation, and the creation of equal opportunities for entrepreneurs (Dy et al., 2016; Martinez Dy et al., 2018; McAdam et al., 2019). Furthermore, when looking at US-based production, we first note the synergies between US scholars, and their UK colleagues, as a few contributions feature both affiliations in the authors’ list (Lall et al., 2022; Nambisan et al., 2019). Generally speaking, however, US production is characterized by a strong focus on the evolution of DE, its challenges, characteristics, and future prospects (Nambisan et al., 2019; Sahut et al., 2019). Moreover, when looking at the Italian scientific community, we once again note strong synergies between countries, specifically with the US (Ughetto et al., 2019) and France (Filieri et al., 2021). Italy-affiliated production seemingly focuses on the opportunities created by DE ecosystems (Elia et al., 2020), and digital startups (Ghezzi & Cavallo, 2020; Ghezzi, 2019), which is somewhat expected given how the Italian landscape is dominated by small to micro-sized ventures.

Our dataset features a robust variety of contributing authors and affiliations, suggesting that DE research has gathered the attention of several research groups worldwide. Tables 3 and 4 sum up the most prolific authors and organizations. Antonio Ghezzi is the most prolific author in the sample, consistent with Tables 3 and 4, which saw Italy as the third most prolific country. The University of Beira Interior is the most active organization in the field of DE. It is worth noting, however, that Case Western Reserve University features the most cited documents in the sample by Nambisan (2017), in accordance with the previously published bibliometric analysis carried out by Zhai et al. (2022). Thus, when examining Tables 3 and 4, a significant caveat should be considered, namely the difference between overall production in terms of papers published, and overall impact in terms of citations. We focused on overall production since Zhai et al. (2022) had already performed an analysis based on impact, which our sample is consistent with, as the work by Nambisan (2017) appears to be the most influential.

Table 3 Most Prolific Authors on DE
Full size table
Table 4 Most prolific Institutions on DE
Full size table

Delving deeper into Tables 3 and 4, we note how Antonio Ghezzi and Angelo Cavallo developed several studies meant to investigate lean entrepreneurship, and the ways in which business angel groups and venture capital funds affect the growth of new digital ventures. Such research relies on data gathered from the Italian entrepreneurial landscape, which is rather lively, and heavily relies on startups and small businesses to function and develop, thus making it an ideal sample to investigate in terms of DE research. Furthermore, Maura McAdam contributed significantly to the field, by exploring the experiences of women entrepreneurs in digital spaces, whereas Sascha Kraus focused on topics such as smart cities, share economy, and innovative business models for digital ventures.

When looking at the outlets publishing research on DE, we note a variety of journals, ranging from entrepreneurship-focused outlets, to general business and social science outlets. Technological forecasting and social change features the highest number of articles. Table 5 depicts in detail the outlets featured in the sample, in order of articles published.

Table 5 Top 10 journals by number of publications
Full size table

When looking at journal-specific research, we note how the most cited articles featured on technological forecasting and social change take a literature review approach, which is either qualitative or systematic (Rippa & Secundo, 2019; Kaushik et al., 2023; Zaheer et al., 2019). Coherently within the scope of the journal, the articles featured on technological forecasting and social change focus on the innovative aspects of DE, and business development (Cano-Marin et al., 2023). Production featured in the International Journal of Entrepreneurial Behaviour and Research is varied, and ranges from research agendas derived to further develop the field of DE (Kraus et al., 2018) and lean entrepreneurship (Solaimani et al., 2017). Furthermore, records featured in the Journal of Business Research explore the business implications of DE, coherently within the scope of the journal. More specifically, Journal of Business Research articles explore the adoption of novel technologies, namely blockchain and artificial intelligence, for the creation of entrepreneurial opportunities (Chalmers et al., 2021; Paul et al., 2023).

Figure 6 was created in VOSViewer. The tool illustrates the connection between keywords through the logic of co-occurrence. It is a valuable tool for visualizing large data portions and identifying specific clusters. In the case of Fig. 3, we can roughly see the same four clusters that we have identified through LDA topic modeling. The red cluster features keywords connected to business model transformation and illustrates the changes brought about by digital transformation. The green cluster refers to the innovation sphere, as we notice keywords hinting at the future of entrepreneurship, effectuation theory, and innovative start-ups. On top, the blue cluster represents the empowerment theme. We see keywords referring to the gender divide and how DE can provide opportunities for emerging countries due to lower barriers to entry. Finally, on the left side, we see the cluster referencing digital platforms. Here we notice how social media and online ecosystems are referenced, how essential communities are, and how digital technologies have shaped them.

Fig. 6
figure 6

Keyword analysis

Full size image

Discussion

Which themes involving DE are the most established and emerging among scholars?

In order to answer RQ2 and uncover the most established and emerging themes among DE scholars, we used the topics uncovered through LDA analysis and grouped them into four distinct themes. In order to do so, we conducted an extensive qualitative thematic analysis process, by carefully considering each of the topics and triangulating each one with each of the records featured in the sample, through open and axial coding. More specifically, the 17 topics extracted through LDA acted as first order codes, and were then grouped in more comprehensive themes by the co-authors. The results were later consolidated and validated by a panel of experts with expertise on entrepreneurship and digital transformation. The validated results are presented below, as we depict how each topic has evolved over the years, and around which themes, from among the more recently emerging ones, academic discussion is focusing on.

Business model transformation through DE

Digital technologies have profoundly disrupted existing business models in both developed and developing countries, enabling enterprises to alter traditional business models, and formulate new and original entrepreneurial ideas. Generally speaking, enterprises are able to enhance their value chain and develop new business models through digital transformation. However, the challenges connected to digital advancements and, more specifically, the shaping of new entrepreneurial ventures around them, has been debated at length within DE research (Jawad et al., 2021). Regardless of the challenges faced by digital entrepreneurs, scholars agree that digital technologies have the potential to both transform existing business models and create completely new ones (Ammirato et al., 2022). DE scholars have empirically examined a few digitally enabled business models to determine their characteristics and impact. Reuschke and Mason (2022), for instance, have uncovered the nature of home-based ventures, featuring a significant variety of arrangements, as some rely on online sales, while others rely on traditional offline commerce.

When looking at the field through a critical lens, to pinpoint established and emerging themes, we note how digital transformation as a whole is constantly evolving and creating business opportunities for entrepreneurs, thus keeping the field lively, and rapidly changing to keep pace with technological advancements. The ways in which digital transformation has disrupted existing business models has been a core theme of DE literature since its inception (Kraus et al., 2018). However, the growing accessibility of digital technologies and the need for practitioners to be constantly aware of how DE is evolving, will keep the scientific debate alive for the foreseeable future (Vadana et al., 2019; Martinez Dy et al., 2018), as more and more traditional businesses will either be forced to evolve or be replaced by digital counterparts (Sahut et al., 2019; Leung & Cossu, 2019).

DE as an enabler of innovation

Digital entrepreneurs make use of technology to create open innovation communities and networking opportunities, from which they generate value for their venture (Elia et al., 2021). This trend has been exacerbated by recent events, namely the COVID-19 emergency and the disruptive diffusion of digital technologies in businesses, thus promoting more and more entrepreneurs to explore the networking opportunities provided by digital communities (Elia et al., 2020). Starting from this premise, DE research has investigated the linkage between DE and innovation (Tang et al., 2022), with a strong emphasis on open innovation specifically, and on how DE facilitates the formation of regional innovation and entrepreneurial ecosystems (Nambisan et al., 2019; Zahra et al., 2023). Huber et al. (2022), for instance, demonstrated how access to open data stimulates digital entrepreneurial activity, further corroborating the link between entrepreneurship and innovation literature. Further proof comes from the study of Duparc et al. (2022), who highlighted how open source business models fuel open innovation, which in turn generates opportunities for the development of DE.

However, when we look more closely at variables affecting innovative DE activity and success, we notice how scholars have come to somewhat diverging conclusions. For instance, Lall et al. (2022) have stressed the difficulties faced by entrepreneurs when looking for online mentorship. More precisely, while online communities have proven to be effective sources of DE opportunities and open innovation (Tang et al., 2022), establishing a proper mentor–mentee relationship is far from being straightforward. In other words, while external factors such as family and community support have a significant effect on fostering DE activity (Basly & Hammouda, 2020; Upadhyay et al., 2022), establishing long-lasting partnerships and mentoring relationships for the sake of open innovation remains challenging.

Empowerment through DE

An interesting stream that has been developing in recent years is the one concerning the empowering role of DE for nascent entrepreneurs. From a theoretical point of view, digital technology is a significant enabler for entrepreneurial empowerment. While ease of access and accessibility applies to every entrepreneur, women entrepreneurial activity has garnered the attention of DE scholars in recent years. More specifically, scholars have highlighted how the gender divide in entrepreneurship has been progressively closing, also due to the empowerment provided by digital technology. In fact, through DE, women entrepreneurs have shown significantly higher levels of empowerment characteristics, namely in relation to investment decision making, education and personal health (Miniesy et al., 2022). The pursuit of education is particularly relevant in the context of DE, as Shukla et al. (2021) have demonstrated that women possessing digital skillsets are more likely to pursue entrepreneurial activity, thus paving the way for a further reduction in the entrepreneurial gender gap.

Empowerment in DE is, however, a broader topic that is not limited to the gender divide. In fact, as pointed out by Martinez Dy et al. (2018), modern digital ventures require minimal resources to operate, thus becoming prime opportunities for everyone, including socially marginalized people who might not have afforded traditional ventures. The above is particularly relevant in contexts where the growing rate of unemployment is a concern for policy makers, such as the case of Africa depicted by Koomson et al. (2022). The study of Biclesanu et al. (2021) corroborated this perspective and further expanded on it, by investigating the driving forces of DE. Their results show how digital businesses are indeed considered easier to establish compared to traditional ones, yet they also noted how digital entrepreneurs believe the digital environment is and will be essential for the foreseeable future, thus further justifying the shift in paradigm. More specifically in terms of ease of access, Biclesanu et al. (2021) list the low number of employees required, the perceived level of risk and the moderate starting investment as the main driving factors behind DE and its role in emancipating novel entrepreneurs.

Overall, our analysis confirms the findings of Alhajri and Aloud (2023), as while promising, the field of female DE appears still to be at a nascent stage, mostly relying on fragmented contributions featured in non peer-reviewed sources. Our analysis corroborated the general lack of theoretical underpinnings, which may explain why the field is only recently starting to emerge. Rather than being established, the field of female DE appears to be an emerging theme that could potentially welcome newer contributions from a wider range of experts, as thus far only a few have shown a high degree of specialization in the topic. When comparing the field of DE to the broader entrepreneurship literature, we notice how empowerment is a much more emphasized trend (Paul et al., 2023), thus suggesting that it might continue to emerge and evolve in the upcoming years.

Digital platforms

Digital platforms contribute to a significant portion of DE literature. By definition, a “Digital Platform is a shared, common set of services and Information Technology architecture that aids to host complementary offerings, including digital artifacts” (Nambisan & Baron, 2021). Digital platforms provide entrepreneurs with a technological environment in which they can develop their entrepreneurial ideas, reach target audiences more effectively and scale up faster. When examining the stream and its development over the years, Srinivasan and Venkatraman (2018) note a gradual shift in scholarly attention towards entrepreneurs on digital platforms, rather than platform providers. The increased attention received from academics over the years is due to their relevance in the modern economy, as digital platforms have simultaneously disrupted traditional business models, while providing digital entrepreneurs new opportunities to develop. An example of the phenomenon is the surge of customer entrepreneurs, who use digital platforms to advertise products, create and share content, and ultimately receive payments (Park et al., 2021). Once conventionally labeled as end-consumers, customer entrepreneurs are one of the latest examples of the enabling role digital platforms have for DE activity.

When comprehensively examining the literature on digital platforms and entrepreneurship, we note how recent efforts have been made to keep pace with the unrelenting evolution of digital technologies and the consequent surge of new platforms. For instance, Chalmers et al. (2021) investigated the use of distributed blockchain technologies, considered to be the natural evolution of traditional digital platforms in the music industry, such as Spotify and Deezer. Additionally, Park et al. (2021) explored the aforementioned nature of consumer entrepreneurs and how they make use of social media platforms such as YouTube to change the paradigm of end-users, moving from a passive to an active role in the consumption of goods.

Additionally, the academic debate has evolved around the benefits and challenges of platform-dependent ventures. While Chandna and Salimath (2022) highlight the potential posited for value co-creation when companies use digital platforms to engage and learn from customers, Ingram Bogusz et al. (2019) warn of the dangers of the perceived illegitimacy of said platforms. In other words, Ingram Bogusz et al. (2019) stress the importance of legitimacy when building a venture on a digital platform, as it is the foundation for its development and the benefits it could draw from the platform itself. The study of Liu et al. (2018) complements this line of thought, as it illustrates the importance of networking to strategically leverage competitive advantage in the context of digital platforms.

Overall, when comparing our results with previous studies on digital platforms, we note the need for further quantitative research to expand upon extant empirical studies. For instance, the customers’ perspective is currently underexplored in the digital platforms literature, while it could be a valid choice for further quantitative contributions to the field. As for the emerging themes, Fernandes et al. (2022) stress the importance of the sharing economy and, more specifically, how it connects with products and services offered through digital platforms. A further emerging topic is one of social media platforms (Park et al., 2021). While social media are considered a pillar of DE literature, the proliferation of new sites and apps means the field is constantly evolving, thus keeping the scientific debate lively around emerging technologies and social media sites.

Theoretical lenses of the sample

As pointed out by Perez-Vega et al. (2022), bibliometric research tends to focus more on citation relationship, thus leaving the theoretical lenses of the studies partially unexplored. In an effort to address this inherent limitation of bibliometric research, we conducted an in-depth keyword analysis to investigate the theoretical underpinnings of the records found amid the sample. We discuss the top theoretical lenses in the section below.

Effectuation theory appears to be one of the most influential theoretical frameworks used among the sample. We refer to the definition of Sarasvathy (2001, p. 245) which refers to “processes that take a set of means as given and focus on selecting between various possible effects that can be created with that set of means”, the opposite of causation, which is “processes that take a particular effect as given and focus on selecting the correct means to create that effect”. In the context of DE, Solaimani et al. (2022) used effectuation theory as a theoretical backdrop to investigate digital entrepreneurs' cognitive and behavioral logic. Additionally, Gabrielsson et al. (2022) have extended effectuation theory to include affectual commitment, as an alternative to the decision-making process originally posited by the theory.

Furthermore, institutional theory also appears often in the sample. By definition, the theory postulates that society is built upon both formal and informal institutions (North, 1990). Recent DE research has drawn on institutional theory to investigate how the digital realm affects formal and informal institutions. For instance, Sarkar et al. (2022) noted the importance of informal open-source ecosystems. Similarly, Hansen (2019) has highlighted how a supportive political, economic and social environment can provide fertile ground for the development of digital entrepreneurial activity, through the theoretical lens of institutional theory.

Aside from effectuation theory and institutional theory, the field of DE features a significant variety of theoretical underpinnings, thus testifying to a somewhat multidisciplinary approach in its research. For instance, Chalmers et al. (2021) drew on external enabler theory to investigate how digital entrepreneurs learn from external enablers, and how they incorporate the lessons learned in their entrepreneurial activity. In addition, Liu et al. (2018) has shown through structural holes theory how information asymmetry amid guanxi networks in China lead to brokers gaining unfair control and information benefits, thus posing some risks when entrepreneurs engage with digital platforms to build their businesses.

Conclusions

This bibliometric research aims to answer a series of RQs formulated by previous studies. Given the scattered and multidisciplinary nature of prior literature on DE, our study has attempted to synthesize extant knowledge and provide an extensive overview of its development over the past years. In doing so, we address the call for a bibliometric review of DE literature by previous scholars, specifically by providing bibliographic and qualitative information on the field and its evolution. While previous attempts have focused on specific aspects of DE, such as digital platforms or adjacent literature streams, such as technological entrepreneurship research, ours is the first attempt to structure recent DE literature comprehensively.

By reviewing the field of DE, we find notable differences compared to traditional entrepreneurship literature, which leads to DE having the potential to significantly disrupt entrepreneurship theory, and bring new and unique perspectives to extant knowledge. Over the past decades, the development of DE as a literature stream has attracted the attention of hundreds of journals and thousands of authors. Our bibliometric review shows the proliferation of the field, as well as the potential it has for further growth, given the vast range of future research directions extracted from the sample, from both a methodological and a thematic perspective.

Our study presents several contributions to the existing body of knowledge on DE and, more broadly, the field of entrepreneurship. First and foremost, our study answers the call of Paul et al. (2023), in regard to the need for the field of DE to be scrutinized under the perspective of a bibliometric review. A further contribution to the field comes from the methodological design implemented in the research as, to the best of our knowledge, no study has used LDA topic modeling to map the field of DE through the use of machine learning and automated text mining, as suggested by Zhai et al. (2022). In doing so, our study draws upon and complements previous bibliometric works (Yu et al., 2022; Contreras Cruz et al., 2022; Gil-Gomez et al., 2021), by examining a field that is yet to be explored bibliometrically through automated, LDA topic modeling.

Implications for theory

As briefly discussed above, our study features several contributions to both theory and practice. From a theoretical perspective, our study provides a clear, comprehensive snapshot of the current field of DE and a set of future research directions that could aid scholars in further developing this area. Our results have highlighted several methodological and empirical research gaps, thus providing authors with a comprehensive research agenda that could incentivize future contributions to the field. Additionally, the in-depth research profile provided could aid scholars in better understanding the current scientific landscape, thus allowing them more effectively to position their research towards publication outlets that might be the most interested in their work. A further theoretical contribution is related to the research design employed in the present analysis. While a few attempts have been made to synthesize previous research on DE (Fernandes et al., 2022; Majdouline et al., 2022; Zhai et al., 2022), no attempt has been made yet to use LDA topic modeling to perform a bibliometric analysis of the field. In doing so, we answer both the call of Paul et al. (2023) regarding performing a bibliometric mapping of the DE field and one from Zhai et al. (2022) regarding employing a novel approach, never applied, thus contributing to the field with a new and unprecedented perspective on its development.

More precisely, a primary contribution of our work is the definition of a research profile, which was thoroughly investigated when we addressed RQ1. In fact, the scope of RQ1 is to provide a comprehensive overview of the DE literature, specifically by analyzing key bibliometric information to determine the most prolific and influential contributors to the field. While the stream has benefited from a significant variety of authors and affiliations, our bibliographic analysis has illustrated who the key authors in the field are, the most prolific institutions, the most influential countries and the collaborative network that has been established around the topic. In doing so, we answer the call made by Paul et al. (2023) and have investigated the key scholars, countries of research, scholar affiliations, and key themes through a bibliometric approach, thus providing scholars with an up-to-date overview of the field which could potentially aid them in better positioning their future contributions.

A further theoretical implication is tied to RQ2, which enquired into key thematic areas that DE literature has focused on over the years. In an effort to provide a unique perspective on the field, we have performed a comprehensive structural analysis of DE research through the use of topic modeling. The LDA approach has yet to be performed within the DE literature, despite the important advantages it offers, namely transparency, replicability and objectivity. Consequently and to the best of our knowledge, our study is the first application of topic modeling to structuring DE knowledge, thus providing originality to the research design.

When looking at the thematic areas extracted through LDA topic modeling, we note their strong relevance in the recent literature and hypothesize that they will characterize the future development of the field as well. For instance, the study by Chalmers et al. (2021) investigating the use of distributed blockchain technologies shows how one of the thematic areas we have extracted, namely digital platforms, will continue to evolve in the near future. Furthermore, we have discussed the topic of business models and the changes brought about by digital technologies and the importance of adopting an “agile” culture within digital ventures, in order to make the most of such changes (Griva et al., 2023). In addition to the variety of models discovered in recent years, we are confident several others will stem from open source collaborations among digital entrepreneurs in the coming years (Duparc et al., 2022; Lin & Maruping, 2022). A third emerging thematic area is tied to the empowering role of DE, as we have seen how lower barriers to entry stimulate entrepreneurial activity. Finally, we have investigated the impact of digital platforms on DE, their logic and their challenges.

Finally, a further theoretical implication is the development of a structured research agenda. By addressing RQ3 in Section “Future Research Directions” below, we have attempted to bridge a known limitation of bibliometric studies, which is the lack of in-depth analysis of the articles featured in the sample, and the consequent lack of a structured research agenda derived from such analysis. In doing so, we have carefully analyzed each of the emerging topics extracted from the sample through LDA topic modeling, and reviewed them from several perspectives, to highlight their content, their methodological design and their theoretical underpinnings. This process has allowed us to identify several opportunities for future research to expand upon the limitations of the present sample, which we have listed earlier. Our intention is that such research directions would aid researchers in developing new contributions to further expand the field of DE in the years to come.

Implications for practice

From a practical point of view, our study contributes in a broader sense as the application of machine learning and, more specifically, LDA algorithms is of evident, practical importance for entrepreneurs and practitioners alike. By providing a structured look at the topics of this study field and showing their development over the years, we allow entrepreneurs to become familiar with scientific production, thus helping narrow the gap between academia and industry. Additionally, our study illustrates several implications of DE for entrepreneurs, namely how it constitutes a significant enabler for empowerment, how it helps reshape their existing business models in light of the changes happening amid the competitive landscape, and how it fosters innovation for their venture.

More specifically, our study provides a practical contribution to female entrepreneurship by highlighting the need for a bridge between practitioners in the field and DE scholars. Thus far, female digital entrepreneurs are somewhat underrepresented in DE research, as scholars have only recently begun to explore their perspectives. Thus, our study encourages establishing a dialogue between practitioners and scholars to give the field of female DE more focus and depth, drawing from practical experiences, success stories, and unique case studies.

A further practical implication comes with DE's implications for technology ambidextrousity, most notably in correlation with the new avenues for research opened by digital platforms. Our study highlights the importance of knowledge ambidextrousity and technological ambidextrousity, as the correlation between the two could help digital entrepreneurs overcome operational barriers and get the most out of digital transformation. Throughout our review, we have seen how digital technology can enable enterprises to diversify their scope, thus potentially allowing them to explore new solutions through the creation of new digital startups while simultaneously exploiting technologies and different knowledge sources to overcome challenges in established businesses.

Finally, a further practical contribution of the study is a better understanding of the distinguished characteristics of DE, in contrast with traditional entrepreneurship. As progress is inevitable, our study could serve as a conceptual map for entrepreneurs to aid them through their digital transformation journey. More precisely, the content of this extensive bibliometric analysis helps practitioners identify the key opportunities, challenges, and critical success factors of DE. In other words, our work strives to be a tool to update the conceptual understandings of practitioners and aid them in developing their digital ventures.

Future research directions

The present section strives to find an answer to RQ3 and outline a structured research agenda for the field of DE. Some of the future research directions discussed below were derived from gaps identified in the selected pool of articles, while others were identified by examining the research design featured in the studies. This procedure has allowed us, for instance, to highlight studies that suffered from replicability issues due to their qualitative and exploratory nature with a limited sample size, thus prompting a call for future studies to gather additional data, and confirm or refute their preliminary findings. Additionally, we derived future research opportunities by examining the research profile of the sample and, more specifically, its geographical distribution. In fact, as noted earlier in the discussion, evidence from emerging countries and Asian regions is relatively smaller in size when compared to the data from the US or Europe. Given how valuable multiple perspectives are, we have thus suggested that future research should undertake a cross-cultural approach by investigating the realities that have yet to receive full attention from scholars. We provide a more structured look at future research directions in the sections below.

Business model transformation through DE

When examining the sample of research on business model transformation, a common limitation is related to the research design. In other words, several studies undertook a qualitative, exploratory approach with a limited sample size (Chandna & Salimath, 2022; Duparc et al., 2022), thus prompting a call for future research to replicate these studies in a different context to confirm or refute their results. Similarly, a qualitative approach could be taken to explore matters already covered through quantitative means. For instance, a qualitative inquiry could explore the lessons learned from COVID-19 and the effects the pandemic has had on the transformation of digital business models (Biclesanu et al., 2021), or the reasons as to why home-based business models are not particularly common in rural areas (Reuschke & Mason, 2022).

DE as an enabler of innovation

Several lines of future research can be drawn when looking at the interlink between DE research and innovation. Broadly speaking, Basly and Hammouda (2020) note that DE is a context-influenced phenomenon, thus future research could expand upon extant empirical studies and replicate them in different contexts, such as different countries, industries and stages of entrepreneurial development, to name a few. For instance, Huber et al. (2022) mentioned how studying the effects of open data on innovation and DE could be valuable for future developments in the field. Thus far, researchers have explored the perspective of nascent entrepreneurial activity, leaving larger institutions and enterprises still partially unexplored under the lens of open data and innovation.

Furthermore, more research is needed to investigate the innovation ecosystems surrounding digital entrepreneurs (Sussan & Acs, 2017). More specifically, the industrial and territorial anchoring of digital entrepreneurs is still relevant, albeit its impact on the development of digital ventures is yet to be fully understood in research. Basly and Hammouda (2020) call for future research on the matter, especially in regard to family-owned enterprises. While most researchers agree on how openness to the surroundings is beneficial for innovation, Lin and Maruping (2022) warn of the potential conflict arising from the digital ecosystem, which could inspire future studies to undertake a more critical approach to the topic. The above is especially relevant in light of the contrast between formal and informal networking, which is also a topic which could be explored by further research (Soluk et al., 2021). In fact, as Lall et al. (2022) also point out, observational or experimental research could be pursued on the topic of the development of networks among digital entrepreneurs, to determine how they are established and what their implications for innovation are.

Empowerment through DE

Empowerment and gender diversity are becoming increasingly popular topics amid DE research. However, while publications investigating them are quickly growing in number, several research opportunities still remain. First, while DE contributes to the tightening of the gender gap, some concerns remain amid the digital ecosystem. As Kang (2022) points out, further research is needed to understand what causes women entrepreneurs to underperform in high-tech environments. Similarly, Jawad et al. (2021) point out that gendered contrasts in entrepreneurship are yet to be fully explored. In other words, there’s a need for empirical, experience-based research to understand the conditions female entrepreneurs experience in a digital environment.

Additionally, the concept of skill development and training is worthy of consideration for future research. More specifically, future research could explore how skill development and training could mediate the gender gap in terms of performance (Kang, 2022; Shukla et al., 2021). The same applies to the use of technological tools, such as social media (Miniesy et al., 2022). A different perspective could take into consideration additional demographic factors (Biclesanu et al., 2021), moving past the gender divide and exploring the differences amid specific categories of digital entrepreneurs.

Digital platforms

Given the unique challenges provided by digital platforms, scholars have called for future research to investigate perspectives. Srinivasan and Venkatraman (2018), for instance, mentioned that digital platforms, as ecosystems, are inherently different from traditional entrepreneurial networks of relationships, thus requiring a distinct set of theoretical lenses and empirical contributions to investigate them. A similar sentiment is echoed by Nambisan and Baron (2021), who noted how entrepreneurs on digital platforms are both users and business owners, thus generating a hybrid figure that could generate stress and other negative effects. While their study has already shed light on the phenomenon, future research could investigate the presence of moderating variables affecting the correlation mentioned above.

In a similar vein, several critical perspectives have yet to be fully explored when discussing digital platforms. For instance, the issue of legitimacy and how it can negatively affect an entrepreneurial venture is yet to be fully understood (Ingram Bogusz et al., 2019), thus making the topics of platform legitimacy and legitimacy building worth exploring further in the future. Additionally, while we have seen how DE has the potential to narrow the gender divide, Kang (2022) highlights how future research could investigate the reasons as to why female entrepreneurs tend to underperform when advanced technological skills are required to operate on a specific platform. Alternatively, more empirical research could refute their claim or provide further validity to it.

Limitations and concluding remarks

We conclude our work by listing a few limitations concerning the methodology. More specifically, the LDA approach features a series of intrinsic limitations that need to be kept in mind when interpreting the results. First and foremost, while LDA is suitable for processing large bodies of research, it cannot provide the same level of depth offered by systematic literature reviews (Kaushik et al., 2023; Ligorio et al., 2022; Wang et al., 2022). Additionally, LDA relies on interpretation by the authors, thus leaving room for human error in the extraction and analysis of the topics. In an attempt to address this limitation, we have performed a qualitative and in-depth review of the possible pathways for future research featured in the sample, in addition to an extensive overview of the theoretical lenses used in DE literature. Since scholars tend to look for the above two elements when engaging with a systematic literature review, our intent was to complement our bibliometric analysis with information that could be of interest to future DE researchers. However, a follow-up study could possibly look individually at each of the topics we have highlighted through a systematic scope, to provide a more detailed look compared to the one featured in the present bibliometric study.

A further limitation is related to the sample and the tools used for the analysis. We collected data in January 2023, thus excluding articles published in 2023 and limiting our scope to those published until December 2022. Given the significant growth in terms of scientific output around the topic of DE, soon more studies will be published, thus paving the way for bibliometric or systematic reviews in the future. Additionally, the use of VOSViewer can be considered a limitation of our study, as other tools such as BibExcel and Sci2 exist. Our choice was due to the compatibility of VOSViewer with WoS, which allowed us to directly import the sample of records extracted from the online repository.

In conclusion, we believe our study could be a starting point for future development of the DE literature stream. As we have comprehensively looked at both its past and present, we noticed how the field is extremely lively and rapidly developing, thus prompting optimism when looking at its future. Consistently with the above statement, we hope our study can provide some guidance on how future researchers could further develop the DE stream and use LDA topic modeling in bibliometric research.