2.1.1. Text-Mining Approach

Academic discourse continues to embrace and integrate technological advances to refine its methodology. One such method, text mining, facilitates the extraction of meaningful insights from extensive unstructured text data through techniques that span natural language processing (NLP), information retrieval, and machine learning (Feldman & Sanger, 2007). This method is crucial for identifying prominent patterns and gaining insights into document corpora. As a specialized realm within text mining, topic modeling, an unsupervised learning technique, is used to identify subjects within a collection of documents (Blei et al., 2003). Unsupervised learning refers to the process in which an algorithm learns patterns from untagged data without any explicit instruction on what patterns to find, making it well-suited for discovering hidden thematic structures in large textual datasets (Hoffman et al., 2010). This methodology has recently garnered significant attention, and has been applied in various academic disciplines.

Widespread recognition of the potential of topic modeling is evident in existing studies, which highlights the complexities of individual research topics within business studies and illuminates interdisciplinary research possibilities. Previous findings significantly underpin our extensive topic modeling approach in previous studies on technology. Moreover, Lee and Bozeman (2005) emphasized the need to understand international research trends in academic journals. Their approach motivated us to survey academic transformations from a global perspective, using the WoS International Academic Database. Although Börner et al. (2003) aimed to offer the academic directionalities required in actual fields, our study pivots more towards directly utilizing topic modeling rather than visualization.

Building on these foundational studies, our study aims to systematically explore changes in academic topics over time through extensive topic modeling of academic databases. These endeavors are expected to shed light on the trajectories of academic development in the past and present, further aiding the forecasting of future research subjects and directions. With the massive amounts of data generated in our digital era, as emphasized by Hilbert and López (2011), effectively analyzing and utilizing these data has become paramount. From this standpoint, our study analyzed more than 7,000 academic studies recorded in 21 Financial Times 50 listed journals via topic modeling to excavate their inherent value and insights. Our research relies on insights and knowledge gained from previous studies and anticipates that the outcomes will provide valuable material to researchers across various fields (Chen, 2006). By evolving and building on the results of prior studies, our research offers robust methodologies and approaches for visualizing academic subject transformations over time and illustrates how they can be harnessed to predict future research trends.

2.1.2. Data and Preprocessing

Data were collected from the WoS database based on 21 journals, including the Academy of Management Journal (AMJ), Administrative Science Quarterly (ASQ), Human Relations (HR), Human Resource Management (HRM), Information Systems Research (ISR), Journal of Applied Psychology (JAP), Journal of Business Ethics (JBE), Journal of Business Venturing (JBV), Journal of International Business Studies (JIBS), Journal of Management (JOM), Journal of Management Information Systems (JMIS), Journal of Management Studies (JMS), Management Science (MS), MIS Quarterly (MISQ), Organization Science (OS), Organization Studies, Organizational Behavior and Human Decision Processes (OBHDP), Research Policy (RP), Sloan Management Review (SMR), Strategic Entrepreneurship Journal (SEJ), and Strategic Management Science (SMJ), and amassed 7,000 scholarly articles from 1983 to 2022. These articles were sourced from various academic journals, including 1,655 from Research Policy and 734 from MIS Quarterly. The selection of these 21 journals, which predominantly focused on management, strategy, and technology, was deliberate. This choice aligns with our research objective of examining the evolution of technology-related studies and their impact on organizational employees. By focusing on journals that closely interlink technology with management and strategy, we ensured that our review comprehensively covered the multifaceted ways in which technology intersects with organizational dynamics and strategic decision-making. This approach facilitates an inclusive glimpse into diverse academic development trends by ensuring a comprehensive scope within the stipulated timeframe.

Topic modeling, a statistical model employed to discern topics within a collection of documents, has been used to analyze the evolution of academic subjects (Blei et al., 2003). Given its recent surge in popularity across various academic disciplines, this is a powerful, unsupervised learning technique. To implement the Latent Dirichlet Allocation (LDA) model, an integral tool for this research, we utilized the Python library tomotopy. The LDA operates on the premise that each document comprises multiple topics, and learns to associate distinctive word distributions with each topic (Blei et al., 2003). The model underwent 200 iterations, with the log-likelihood values produced at each stage to provide real-time insights into the model’s performance enhancement. The top 10 words from multiple topics were extracted to discern the essence of each topic.

The preprocessing phase is crucial for data refinement. Using the natural language toolkit (nltk) package, a comprehensive suite of libraries and programs for symbolic and statistical NLP in the English language, we retained English stems and excised extraneous elements such as special characters. The nltk is widely used in text mining for tasks such as tokenization, stemming, and tagging, making it an appropriate choice for ensuring the accuracy and consistency of linguistic analysis. Additionally, English stop words were excluded to ensure a focus on core semantic content (Bird et al., 2009). To achieve a granular analysis of data from 1983 to 2022, the research duration was segmented into four distinct intervals: 1983-1992 (Time period 1), 1983-2000 (Time period 2), 1983-2010 (Time period 3), and 1983-2022 (Time period 4) to see how studies have evolved. The use of overlapping periods was integral to our approach, allowing us to observe the cumulative and evolving nature of this research. By separately examining periods such as 1983-1992 and 1993-2002, we identified distinct trends within each decade. However, overlapping periods are necessary to fully understand the evolution and additive impact of research over time. For instance, examining 1983-1992 and then extending this to include 1993-2000 reveals not only the continuation of trends from the first period, but also the emergence of new ones, providing a clearer picture of the study’s evolutionary trajectory. By doing so, we can trace evolutionary paths and key topics by investigating trends with overlapping periods. Topic modeling was executed for each segment, enabling a structured understanding of the evolution of academic subjects over time. After the model training, the primary themes were interpreted using the top words associated with each topic. These interpretations are underscored as one of the principal outcomes of this study.

2.1.3. Exploratory Data Analysis and Optimization

Embarking on the preprocessed data, this study applied an LDA topic-modeling algorithm. Determining the most appropriate number of topics is crucial for modelling precision and efficiency. The coherence and perplexity metrics map the evolution of academic research topics over time. A grid-search method was used to identify the optimal parameter values during the model learning phase. This grid encompasses Alpha and Eta values, considering cases such as [0.01, 0.05, 0.1, 0.2, 0.3] and topic number k as [5, 8, 10, 12, 15, 20, 25, 30]. After training the LDA model for each combination, coherence and perplexity values were computed, leading to the selection of the best parameter combination (Röder et al., 2015). Coherence, which exemplifies the congruence of words within a topic, indicates optimal performance with higher values. By contrast, perplexity shows the model’s capability to predict new data; lower values indicate more desirable outcomes (Newman et al., 2010). The decision to use 10–20–20–30 topics across the four phases was based on the results of the grid search (Table 1). We aimed to balance the granularity of topics with the overall coherence of the models. As the volume and complexity of the data increased over time, progressively larger topic numbers (from 10 in the earliest phase to 30 in the latest) offered the best balance between detailed topic representation and model coherence.

In the LDA learning phase of our study, we used an optimal combination of parameters to ensure the high reliability of topic modeling. This approach allowed us to focus on the evolution of research subjects as reflected in the incremental increase in the number of topics discerned from the optimization process. This increase from 10 to 30 topics across the four periods (1983-1992, 1983-2000, 1983-2010, and 1983-2022) mirrors fluctuations in technological and sociocultural trends. In evaluating these topic trends, we deliberately chose not to consider topic frequencies. This decision was made to avoid potential biases arising from the disproportionate representation of certain topics in the larger datasets. Hence, our focus was on the diversity and evolution of topics rather than their frequency. Regarding the data presented in Table 1, a direct comparison of different time periods with varying data point sizes (24, 1,177, 3,495, and 7,000) presents a unique challenge. The stark contrast in data-point volumes across these periods necessitates a nuanced approach to analyzing and interpreting the evolution of topics. This is particularly evident when juxtaposing the initial period (1983-1992) with subsequent periods, as the expansion in data points reflects not only the growth of the field, but also the broadening scope of the topics explored.

One of the pivotal observations of this study was the rapid proliferation of academic papers over time, as depicted in Fig. 1. This surge in data led to the decision to extract 10, 20, 20, and 30 topics across the delineated time intervals, respectively. This approach ensures that the primary topics of each period do not merely represent the themes emerging during that epoch, but more crucially, reflect the relative frequency of their coverage in academic discussions. By adopting this strategy, this study transcends the mere chronological mapping of topic evolution and successfully captures fluctuating academic interests across different timeframes.

Fig. 1

Number of publications over time.

2.1.4. Results and Interpretation of Topic Evolution over Time

Diving into topic modeling outcomes for each period yields the following insights (Table 2):

Time period 1 (1983-1992): During this nascent phase of technological advancement, the primary emphasis was on “Information Technology Management and Problem-solving.” This highlights the initial stages of computer technology and the emergence of the Internet, in which research was chiefly centered on optimizing IT resource management and devising innovative problem-solving methodologies using these technological instruments.

Time period 2 (1983-2000): This interval heralded a more diverse scholarly landscape, encompassing subjects such as “Research Methodology and Performance Measurement Analysis,” “Technological Development and Organizational Information Systems,” and “Technology Innovation and Industry Adoption Dynamics.” This period reflected academia’s proactive response to swift technological advancements, emphasizing the intersection of technological development and organizational management methodologies.

Time period 3 (1983-2010): During this period, salient themes emerged, such as technological innovation, industrial policy and research models, and information system management. Simultaneously, the emergence of themes such as “Team Management and Project Development Performance” and “Organizational Change and Social Theory in Technology” emphasized the growing confluence of organizational behavior and technological breakthroughs. Such themes underscored the increasing significance of technological policies, management of information systems, and the nuanced dance between organizational metamorphoses and technological progress.

Time period 4 (1983-2022): The most recent scholarly trends gravitate towards “Firm Performance and Investment Value,” “Technology Policy and Innovative Science,” and “Research Models and Technology Development.” Concurrently, subjects such as “Product Pricing and Market Dynamics in Technology” and “Organizational Learning and Knowledge Management” have garnered significant academic attention. This indicates a sophisticated evolution in research priorities from preliminary technological management considerations, to a holistic evaluation of organizational performance in the face of technological advancements and their strategic implications.

In summary, topic modeling elucidates the discernible evolution in research emphases across various epochs. From foundational themes centered on IT management at the dawn of the technological age to the intricate considerations of technological policies, organizational performance metrics, and investment implications in contemporary times, this evolution underscores academia’s dynamic response to the symbiotic relationship between technology and management. Such shifts reiterate the critical role of advanced analytical methods, such as topic modeling, in shaping the trajectory of scholarly endeavors, ensuring that they remain agile and pertinent in a perpetually evolving technological and managerial milieu.

Academic research’s beauty lies in its ability to trace the ebb and flow of intellectual curiosity over time. Examining the evolution of specific topics in our study makes it evident that the scholarly focus has shifted and transformed, echoing larger societal and technological changes. As an illustration, Topic 7 from Table 2 (1983 to 1992) was dominantly classified as “Information Technology Management and Problem Solving,” emblematic of a nascent period when IT and problem-solving management were of prime interest. However, this topic morphed into broader and more sophisticated arenas as decades passed. By 2000, it transmuted to Topic 12, emphasizing “Technological Development and Organizational Information Systems,” and by 2010, it had evolved yet again to Topic 3, shedding light on “Team Management and Project Development Performance.” This dynamic shift from a focus on technology management to the nuanced aspects of team dynamics and project success within organizations signals a maturation in academic pursuits, culminating in 2022 with Topic 6, “Decision-Making and Information Control Systems.”

Such mutable narratives of a single topic encapsulate how academic interests are not static but reflective of broader societal, technological, and organizational shifts. This is further illustrated by the Sankey Diagram (a powerful visualization tool widely used across sectors such as marketing and energy analysis), as shown in Table 2. While the volume of data for the period up to 1992 was deemed insufficient for comparison, the visualization from 1983 to 2000 elucidates the flux in popular topics over time, offering a holistic overview of interconnected academic trajectories.

Fig. 2 provides a panoramic view of the flow of popular topics through 2000, 2010, and 2022. It meticulously traces the progression from micro-level insights such as “Product Development and Strategic Manufacturing” to macroscopic evaluations focusing on “Investment Value and Performance Measurement in Technology” and, eventually, “Firm Performance and Investment Value.” This trajectory mirrors a broader shift in research emphasis from an initial focus on tangible product development and operational efficiency to a broader, long-term perspective on organizations’ overall value and performance in the technological landscape.

Fig. 2

Evolution of important topics over time.

Moreover, the transformation from “Technological Development and Organizational Information Systems” to themes such as “Organizational Change and Social Theory in Technology” and “Organizational Learning and Knowledge Management” underscores a nuanced evolution. Organizations have transcended merely altering their internal processes and information systems to strategically position themselves within larger societal narratives. The elevated emphasis on continuous learning and knowledge management foregrounds the importance of technological advancements and accentuates the intricate interplay between intra-organizational dynamics and external technological changes. This intricate interplay highlights the importance of management research for technology-related studies. Understanding its ramifications for organizational structures, processes, and strategies becomes paramount as technology continues its inexorable march forward. Thus, management research bridges technological advancements with organizational realities, ensuring that innovations are harnessed effectively, ethically, and strategically.

In the nexus between interpretation and introduction of new technologies, the increasing centrality of technology’s role in shaping workplace dynamics becomes evident. The interpretive journey from initial technological product development to its current profound influence on organizational value and performance emphasizes the intrinsic relationship between technological advancement and management. Organizations have evolved from viewing technology as a tool for operational optimization to recognizing its fundamental role in shaping organizational structures, processes, and long-term strategies. Such an evolution parallels scholarly recognition that the successful adoption of new technologies hinges significantly on employees’ willingness and capability to integrate these technologies into their work roles and practices (Blanka et al., 2022; Eller et al., 2020). The introduction and adaptation of new technologies in the workplace are not solely a function of the technologies but are interwoven with the broader organizational fabric. As technology becomes an institutional infrastructure component, its effects resonate beyond mere operational facets, shaping the sociotechnical contexts in which employees operate (Barley & Kunda, 2001; Orlikowski & Barley, 2001). This evolving symbiotic relationship underscores the imperative for management scholars to delve deeper into the nuanced interplay between technology and its human and organizational implications. Ensuring that the benefits of technology are harnessed while mitigating potential disruptions requires a holistic understanding of both technological tools and the organizational and individual contexts in which they are situated.