Research on the identification of disruptive technology topics for future industries – taking the field of quantum communication as an example Open Access

The “14th Five-Year Plan” and the “2035 Vision Goals Outline” emphasize the need to “plan and layout a number of future industries” in the field of brain-like intelligence, quantum information and other cutting-edge technologies and industrial changes and strengthen the multi-path exploration, cross-integration and disruptive technology supply of cutting-edge technologies. Future industry is a key emerging industry that aims to meet the future development needs of human society. Driven by disruptive or forward-looking technology, it can reshape global economic and social development models. It has become a core area for developed countries to seize their future competitive advantages. However, at present, there are some problems in the process of future industrial development in China, such as unclear disruptive technology topics and development paths, which limit the development of future industries. Therefore, it can be seen that accelerating the identification of disruptive technology topics for future industries is not only an important driving force for China to achieve the goal of building an innovative country and a scientific and technological power but also a critical pathway for China to compete for the commanding heights of a new round of international competition.

Technology topic identification is the basis for disruptive technology identification. In the early stage, it mainly relied on expert experience to identify technologies in specific fields. It mainly adopts various research methods such as questionnaire surveys and expert consultations. In specific practice, the technology roadmap and the TRIZ method were adopted. For example, Vojak and Chambers (2004) proposed a heuristic method based on historical research cases using a technology roadmap to identify potential disruptive technologies, and Sun et al. (2008) established the necessary and sufficient conditions for predicting innovative technology based on the ideal final result of the product. Based on the theory of technology evolution in TRIZ (theory of inventive problem-solving, creative problem-solving theory, including technology maturity prediction, technology evolution, contradiction resolution, effect, etc.), the prediction process model evolution route of disruptive technology innovation was established to identify potential disruptive technology innovation and a video game system was used as the object for case analysis.

Due to the scarcity of expert resources, it is impractical to analyze massive technical topics and is prone to strong subjectivity. Therefore, it is difficult to meet the needs of disruptive technology topic identification, as many scholars have begun to focus on bibliometric methods and identify technical topics around the characteristics of patent citations. For example, Momeni and Rost (2016) obtained the technology development path based on patent citation data and analyzed the technology development trend through topic modeling to identify disruptive technologies that may achieve subversion. Ma and Wang (2021) used the International Patent Classification (IPC) information to construct the patent similarity index and introduced the time dimension to measure the novelty, uniqueness and influence of patents in the past, current and future three time periods, so as to construct a comprehensive identification scheme for breakthrough technological inventions. Based on summarizing the characteristics of disruptive technology, Li et al. (2021) constructed a disruptive technology identification model from the four dimensions of technology integration, novelty, expansion and influence, based on the patent perspective. The above research methods based on patent citation characteristics overcome the problems of subjectivity, but they often suffer from citation delays, which compromise identification accuracy to some extent.

The above research provides ideas for the identification of disruptive technology topics; however, the topic identification of disruptive technology mostly uses a single data source, and forward-looking technology topic identification is mostly based on patent data. Although patent data contain technical and market attributes, they only represent the practical results of technology and still lack an explicit representation of technology development. As the scientific soil of technology brewing, paper data can provide an earlier basis for the forward-looking identification of technology. Therefore, this study aims to comprehensively use papers and patent data sources to carry out topic identification of disruptive technologies and grasp the frontier trend of science and technology development for future industries from the perspective of scientific research and technology application.

In doing so, this research seeks to address recent calls for a more theoretically grounded and methodologically robust approach to technology foresight. The conceptual ambiguity surrounding “disruptive technology” – its conflation with related concepts like “breakthrough” and “emerging” technologies – has been a persistent challenge in the literature (Qu et al., 2023; Qiao et al., 2022). To overcome this, our framework is explicitly built upon an integrated understanding of disruptiveness, which we conceptualize as a confluence of technological novelty, superiority and positive externality (Qu et al., 2023; Deng et al., 2025). Furthermore, our multi-source methodology aligns with the push for more comprehensive data fusion in knowledge discovery, as successfully demonstrated in other complex domains like public safety and policing strategy analysis (Basilio et al., 2019; Basilio et al., 2021a; Basilio et al., 2021b). By applying this enhanced theoretical and methodological lens to the critical case of quantum communication – a field central to national strategic development (Sun et al., 2025) – this study aims to provide a more precise and meaningful identification of potential disruptive topics.

In recent years, with the rise and wide application of machine and deep learning, an increasing number of scholars have used algorithm models for emerging technology topic identification, including the latent Dirichlet allocation (LDA) topic identification model. For example, Gao et al. (2023) used the LDA topic model to identify hidden technology topics in the field of integrated circuits, and compared and screened emerging technologies by constructing multidimensional indicators such as novelty, intensity and heat.

While the three characteristics of novelty, convergence and growth provide a foundational understanding, a more nuanced delineation of “disruptive technology” is required to distinguish it from adjacent concepts and address the critiques of theoretical ambiguity. Building on the foundational work of Christensen (1997) and integrating recent scholarly discourse, this study refines the conceptual framework by first distinguishing disruptive technologies from related paradigms and then elaborating on its core dimensions.

The terms disruptive technology, breakthrough innovation and emerging technology are often used interchangeably, yet they encapsulate distinct conceptual nuances. Clarifying these differences is essential for a robust theoretical foundation (Qu et al., 2023):

• Disruptive technology vs breakthrough technology. While both concepts imply a significant departure from existing norms, their primary emphases differ. Breakthrough technology (or radical innovation) is predominantly defined from a technology perspective, focusing on the novelty of the knowledge base and a fundamental shift in the technological trajectory (Tushman and Anderson, 1986). It concerns the “newness” of the technology itself. By contrast, disruptive technology is inherently defined by its market perspective and its eventual impact. A technology, even if not a radical breakthrough, can be disruptive by entering a niche market and gradually displacing an established incumbent (Christensen, 1997). As Qiao et al. (2022) articulate, a breakthrough innovation can be seen as one form of disruptive innovation, but not all disruptive innovations stem from a technological breakthrough; they can also arise from the novel application of existing technologies.

• Disruptive technology vs emerging technology. An emerging technology is characterized by its novelty, rapid growth and potential to exert a substantial economic and social impact in the future (Rotolo et al., 2015). It is a technology in its nascent stage, where its ultimate trajectory and impact are still highly uncertain. Disruptive technology, therefore, represents a specific outcome or potential of an emerging technology. All disruptive technologies are, by definition, emerging at some point, but not all emerging technologies will follow a disruptive path; many may follow a sustaining trajectory or simply fail to gain market traction (Qiao et al., 2022). The key differentiator is the proven or potential capacity to reshape market structures.

To capture the multifaceted nature of disruptive technology, this study adopts an integrated framework that synthesizes its core attributes from multiple perspectives. Building on the work of Qu et al. (2023) and Deng et al. (2025), we conceptualize a disruptive technology through three indispensable dimensions: technological novelty, technological superiority and technological externality:

1. Technological novelty (the precondition). Novelty is the foundational prerequisite for disruptiveness (Strumsky and Lobo, 2015). It signifies a fundamental departure from the principles, components or recombination patterns of existing dominant technologies. This “newness” can manifest as an original innovation based on new scientific principles, or as a novel integration and application of existing technologies. While novelty alone does not guarantee disruption, a technology lacking novelty is, by definition, sustaining and therefore incapable of being disruptive. This aligns with the concept of radicalness and breakthrough from a purely technological standpoint.

2. Technological superiority (the competitive driver). Novelty must be coupled with superiority to drive adoption and eventual market dominance. This superiority is not merely about being different; it is about providing a compelling competitive advantage, which can be realized in two primary forms (Schumpeter, 1942; Suarez, 2004). The first is functional superiority, where the new technology enables performance attributes that the dominant technology cannot, eventually outperforming it on key metrics valued by the mainstream market (e.g. digital photography surpassing film). The second is cost superiority, where the technology can deliver comparable or adequate functionality at a significantly lower cost, enabling a low-end market disruption strategy (Christensen, 1997).

3. Technological externality (the enabler of legitimacy and diffusion). A critical and often overlooked dimension is the technology’s externality – its impact on stakeholders beyond the immediate producers and consumers (Roper et al., 2013). A technology’s potential for widespread adoption and diffusion is significantly enhanced by its positive externalities, such as contributing to social welfare, environmental sustainability or national strategic goals. Technologies with negative externalities, regardless of their novelty or superiority, face significant barriers to legitimation and are less likely to achieve large-scale diffusion (DiMaggio and Powell, 1983). For instance, in the context of China’s development strategy, technologies that enhance “new quality productive forces” and contribute to national security are more likely to receive policy support and resource allocation, accelerating their development and deployment (Sun et al., 2025).

By integrating these three dimensions, this study moves beyond a simple listing of characteristics to propose a more robust conceptual model. A technology is truly disruptive only when it successfully combines technological novelty, a path to competitive superiority and a positive externality that aligns with broader social and strategic needs. This integrated framework provides the theoretical lens for our subsequent empirical identification of disruptive technology topics in the field of quantum communication.

The LDA model, as proposed by Blei et al. (2003), serves as the core algorithmic foundation for this study. While acknowledging the emergence of more recent methodologies such as dynamic topic models (DTM), structural topic models (STM) and transformer-based approaches like BERTopic, the selection of the classic LDA model is a deliberate and theoretically grounded choice, justified by the specific objectives and constraints of this research.

First and foremost, this study aims to conduct a retrospective analysis of a fixed, two-decade corpus (2002–2021) to identify and categorize technological themes that have exhibited disruptive potential. Classic LDA is ideally suited for this task as it provides a stable, static representation of the latent thematic structure within a defined corpus. By contrast, DTMs are designed to explicitly model the evolution of topics over time, which, while valuable for longitudinal studies, introduces additional complexity that is not the primary focus of our current investigation – which is to first establish a clear, time-invariant thematic landscape.

Second, the principle of methodological parsimony and replicability strongly favors LDA. The model’s relative simplicity, well-documented algorithm and availability of robust, validated implementations ensure that our results are transparent and easily replicable by other researchers in the field of technology foresight. More complex models, particularly deep learning-based approaches like BERTopic, often function as “black boxes,” making it difficult to trace the precise reasoning behind topic assignments and to conduct sensitivity analyses on key parameters.

Third, the interpretability of LDA outputs is paramount for our subsequent validation and classification stages. Our framework relies on domain experts interpreting the top terms of each topic to assign meaningful labels. The probabilistic word-topic distributions generated by LDA are highly interpretable, providing a clear and intuitive representation of each theme. While BERTopic can generate coherent topics, its reliance on contextualized embeddings can sometimes produce clusters that are semantically consistent but difficult to characterize with a small set of highly probable keywords, posing a challenge for expert validation. Similarly, the additional structural covariates incorporated in STM, while powerful for specific causal inferences, are not required for our primary goal of theme discovery and classification.

Finally, our research design is specifically tailored to multi-source heterogeneous data fusion, integrating scholarly papers and patents. Classic LDA provides a consistent and unified probabilistic framework for modeling both types of textual data, enabling a fair and direct comparison of the resulting topic structures through similarity measures like JS divergence. The application of LDA to such multi-source knowledge discovery problems has been successfully demonstrated in related domains, such as the analysis of policing strategies and operational demand in law enforcement (Basilio et al., 2019; Basilio et al., 2021a; Basilio et al., 2021b), establishing a proven methodological precedent for this study.

In summary, while acknowledging the capabilities of newer models, the selection of classic LDA is driven by its appropriateness for a retrospective, cross-sectional analysis, its methodological transparency and replicability, the superior interpretability of its outputs for expert validation and its proven efficacy in fusing heterogeneous data sources. This choice ensures that our methodological approach is both rigorous and precisely aligned with our research objectives.

The LDA model is based on the bag-of-words approach to convert a document into a word-frequency vector. Each document selects a number of topics with different probabilities and each topic selects a number of words with different probabilities. The “document-topic” and “topic-feature word” matrices of the text were extracted to reduce the dimension of the data. The specific mathematical description of the LDA model is as follows:

• For each document m∈M, according to θm∼Dir（α）, m∈[1, M], the probability distribution θm of the topics in document m is obtained.

• For each topic z∈K, according to Фz∼Dir（β）, z∈[1, K], the probability distribution Фz of the words in topic z is obtained.

• For the nth word wm,n∈[1, Nm] in the document m, according to the multinomial distribution zm,n ∼ Multinomial（Фz）, the topic zm,n corresponding to the word is obtained; according to the multinomial distribution wm,n ∼Multinomial（θzm,n）, the vocabulary wm,n is obtained.

where M represents the number of documents, K represents the number of topics and Nm represents the number of words in the mth document. α and β are parameters of Dirichlet distribution, which are preset constants.

The topic identification of disruptive technology is an important link for cultivating future industries. The purpose of this study is to identify disruptive technology topics in the field of quantum communication with the help of the LDA topic model to further extract the future development characteristics and development direction of the quantum communication field and release new kinetic energy to further cultivate future industries. This research mainly includes pre-data collection and data pre-processing, the LDA topic model and multi-dimensional index-based disruptive technology screening to jointly identify future technology topics. The empirical process is illustrated in Figure 1.

where Nz represents the novelty of topic z, n represents the number of papers or patents on topic z, yi represents the publication year of the ith paper or the year of patent application.

• Calculate the similarity between topics. There is an interaction between scientific research and technological application topics. Therefore, this study uses the Jensen-Shannon (JS) divergence method to measure the similarity between topics as the basis for determining whether it is a common topic of different data sources. The calculation equation is as follows:

Which $M=12(P+Q)$⁠, D is KL divergence $D(P||Q)=∑iP(i)log(P(i)Q(i))$

where p and q are the probabilities of Zm and Zn in their respective vocabularies, respectively. The JS distance interval is [0, 1]. The smaller the JS distance between the two topics, the greater the similarity is (Li et al., 2014). This study determines whether it is a common topic by setting a threshold, that is, when the similarity between the two topics is less than the threshold, the two topics are determined to be common topics; otherwise, they are determined to be non-common topics.

Based on the research of Bai et al. (2020), this study determines common topics and non-common topics based on topic similarity and determines the type of frontier topics based on topic novelty and intensity. These topics can be divided into the following categories.

Hot research frontier topics. Research topics with high topic novelty and high research intensity appear late but have received much attention. Disruptive technology has the characteristics of novelty and influence, so it is easy to produce in these topics. This study classifies it as a “hot topic” in disruptive technology.

Emerging research frontier topics. A research topic with high topic novelty but low research intensity is a research topic with great development potential that has emerged in the recent period but has not received extensive attention. This is in line with the novelty and uncertainty of disruptive innovation. Therefore, this study classifies it as an “emerging topic” in disruptive technology.

Mature research frontier topics. The low novelty of the topic indicates that it appears earlier, which does not conform to the characteristics of novelty and uncertainty of disruptive innovation. This paper names the topics that meet these characteristics as “mature topics.”

Potential research frontier topics. Non-common topics refer to topics that appear only in one data source. When the topic is highly novel in the data set, it is consistent with the type of disruptive technology that may be generated from new innovation or technological innovation applications. Therefore, this study classifies it as a “potential topic” for disruptive technology.

Mature research frontier topics. When the novelty of the non-common topic is low, it indicates that the topic only appears in one data source and has become a “past tense,” and it is less likely to produce disruptive technological innovation. Therefore, this study classifies it as a “mature topic.”

As a typical representative of the new round of scientific and technological revolution and the future industry, the quantum information industry, in which quantum communication-related technology is the frontier technology of the intersection and integration of quantum mechanics and information technology, is a subversive future industrial technology. With the emergence of new technologies such as big data, the internet, artificial intelligence and the Internet of Things, many new applications and models have been developed, and higher requirements have been put forward for information technology.

Identifying technical topics in this field is of great significance for China’s forward-looking layout of disruptive technologies, future industries and the construction and cultivation of relevant policy systems. Therefore, this study selected the field of quantum communication as the research object and conducted disruptive technology identification based on two data sources: papers and patents.

The paper data used in this study are from the Web of Science (WOS) Core Collection, and the patent data are from the Derwent Innovations Index (DII):

1. Paper data acquisition: On January 28, 2022, the WOS database was queried on the topic of quantum communication and the literature type of article. It was searched in the WOS Core Collection (citation index: SSCI, SCI-expanded library). The time range was 2002.01.01–2021.12.31, and 13,721 data points were obtained. The data included the title of the paper, keywords, author keywords, additional keywords, abstracts, citation frequency and publication year.

2. Patent data acquisition: On January 28,2022, the Derwent Patent Search Library was queried on the topic of (quantum communication) (quantum key) or (quantum channel.) or (quantum entangl+), [quantum and (kd or qt)], (quantum cryptography), (electrified photon), (photon entanglement), (quantum and teleportation), [quantum and (key distribution)], (quantum network), (quantum and teleport+) or (quantum satellite? (quantum relay), (quantum gateway), (quantum random number), (quantum random switch), (quantum switch), [quantum and (single photon)] or [quantum encryption)] . The time range is 2002.01.01–2021.12.31. A total of 15,637 data were obtained, including patent name, abstract, technical focus abstract, equivalent abstract, number of Derwent entry registrations and other information.

The selection of the time range from January 1, 2002, to December 31, 2021, is strategically motivated by the developmental trajectory of quantum communication technology. This two-decade period corresponds precisely to the rapid ascendance and maturation phase of quantum communication as a disruptive technological field. The early 2000s marked a pivotal transition from theoretical exploration to experimental validation and initial practical implementations. For instance, the first demonstration of quantum key distribution (QKD) over optical fibers and free space laid the groundwork for subsequent technological breakthroughs during this era. Throughout this period, the field witnessed several transformative milestones, including the development of practical QKD protocols, the establishment of metropolitan quantum networks and the launch of the world’s first quantum communications satellite (Micius) in 2016. By 2021, quantum communication had evolved into a strategically critical domain with demonstrated real-world applications, making the endpoint a natural juncture for assessing the field’s development. Capturing this entire 20-year window allows our analysis to encompass the complete lifecycle from the field’s infancy through its explosive growth, thereby providing a comprehensive foundation for identifying technologies that exhibit the characteristics of novelty, growth and potential disruptiveness as outlined in our theoretical framework.

Raw collected data often contain redundant information and inconsistent formats. Therefore, data preprocessing operations are required, including field merging, word segmentation, stop word removal and data cleaning of the collected papers and patent data. To express the paper and patent topic more comprehensively, the paper data, article title, author keywords, keywords plus and abstract fields are merged, the merged text content is segmented, and the stop words are removed. The patent data are combined with TI (patent name), AB (abstract), TF (technical focus abstract) and EA (equivalent abstract) fields, and the combined text content is segmented and de-stopped.

To further eliminate the error problems caused by factors such as single and plural words, case, word form transformation and tense that may occur in the data after word segmentation processing, we first perform stem extraction and then perform topic identification during data processing.

First, perplexity is calculated to select the optimal number of LDA topics, and then LDA topic identification is performed according to the optimal number of topics:

• Perplexity calculations

By calculating the topic confusion degree of the pre-processed data, it is iterated that when the K value is 2–15, the calculation results of the topic confusion degree of the paper and patent heterogeneous data sets under different topic numbers are shown in Figure 2. Generally, the smaller the perplexity, the better the model fitting effect. Therefore, the K value corresponding to the inflection point of perplexity is selected as the optimal number of topics. As shown in Figure 2, the maximum degree of distortion of the paper data and patent data was 5.

• LDA topic identification

According to the calculation results of perplexity, the number of topics in the LDA topic model in the study and patent data is set to five, and the α and β parameters of the LDA model are set by default. After topic identification of the two data sets, five topics were obtained, and the specific results are shown in Tables 1 and 2.

• Topic similarity calculation

After obtaining the LDA topic identification results of the papers and patent data, we selected the JS distance to calculate the similarity between topics, and the calculation results are shown in Table 3.

• Research topic strength and novelty calculations

According to the topic intensity and topic novelty measurement formula given above, the topic intensity and novelty index of the identified paper topics and patent topics were calculated, respectively. The calculation results are presented in Tables 4 and 5, respectively.

According to the calculation results and the classification criteria of the different types of innovation frontier topics proposed above, the research topics of the identified papers and patent data sets were compared and analyzed, and the corresponding types of innovation frontier topics were identified. In the process of comparison and identification, according to the existing experience and the actual situation of this paper, the topic intensity is sorted in descending order, and the results ranked in the top 60% are marked as “strong topic,” otherwise marked as “weak topic”; the topic novelty is ranked in descending order, and the results in the top 80% are marked as “high novelty,” otherwise marked as “low novelty”; according to the similarity calculation results of this paper, the data with JS distance greater than 0.67 is marked as “non-common topic,” namely, the topic is only reflected in the single data source of the paper or patent.

According to the above marking method, the comparison results of papers and patent data are shown in Table 6, and the research frontier types corresponding to each topic are shown in Table 7:

• Results analysis

The content of each topic type was analyzed according to the identification results in Table 7. Table 8 presents some of the analysis results.

Hot frontier topics include DII_0 | WOS_3 | DII_3 topics. Through an analysis of the subject words corresponding to this type of topic, it was found that this type of topic mainly focuses on research on optics, photons, light, systems, bases and so on.

Taking WOS_3 as an example, its main content is related to the research of quantum information carriers such as optics, single photons, optical fibers and lasers. In the application of quantum communication technology systems, the form of a quantum channel combined with an auxiliary classical channel is primarily adopted. Although quantum communication technology has gradually achieved breakthroughs, an ideal single-photon source suitable for QKD still does not exist. There have been some problems that cannot be ignored, such as the single-photon separation attack and low single-photon efficiency of photon source generation. Therefore, the optimization of information carriers and technological breakthroughs in quantum communication is a hot topic in this field, and it is prone to disruptive technological innovation.

The emerging frontier topics included the WOS_0 | DII_2 topics. By analyzing the keywords corresponding to this type of topic, it was found that this type of topic mainly focuses on the research of signal, module, unit, electron, dot, field and detect.

Considering WOS_0 as an example, its main content is related to the operating system, model, structure, dynamics, effect, calculation, field and other related research. With the development of emerging technologies such as machine learning and deep learning, and the expansion of their application scope, research in the field of quantum communication has gradually adopted related technologies to optimize, design and even achieve technological breakthroughs in quantum communication technology, which has become an emerging frontier topic in this field. Disruptive technological innovation opportunities also exist.

Potential frontier topics include WOS_2 | WOS_4 | DII_4 topics. By analyzing the keywords corresponding to this type of topic, it was found that it mainly focuses on key, protocol, channel, dot, distribute and other aspects of research.

Taking topic WOS_2 as an example, its main content is related to key distribution, security protocol, quantum state, channel and research related to QKD protocols. Security is an important advantage of quantum communication technology. Research on related topics, such as communication protocols and information security assurance, is important for realizing quantum communication applications. Therefore, the development of new and effective communication protocols to ensure information security is a potential research focus in this field. This belongs to the field of potentially disruptive innovation in the future.

The mature frontier topics include WOS_1 | DII_1 topics. By analyzing the keywords corresponding to this type of topic, it was found that this type of topic mainly focuses on the research of state, entangle, atom, scheme, teleport, spin and semiconductor.

Taking the topic WOS_1 as an example, its main content is related to quantum state, entanglement, atom, photon, scheme, transmission, spin and research related to quantum physics, namely, the physics-related basic research of quantum teleportation. As the theoretical basis of quantum communication, photons, atoms, quantum entanglement, entanglement effects, quantum spin states, etc. are becoming increasingly widely used with the deepening of research and demonstration. Based on these mature research foundations, future research will seek to discover and solve new problems and further promote the innovation and development of scientific research and technological applications.

In this study, natural language processing technology and the LDA topic model were applied to research on disruptive technology topic identification. Taking the field of quantum communication as an example, this study selected papers and patent heterogeneous data sources, determined common and non-common topics based on topic similarity and identified disruptive technology innovation topics based on topic novelty and topic strength. It provides a scientific reference for enterprises to accurately and efficiently conduct disruptive technology R&D and accelerate the cultivation of future industries. The study found that:

• Multi-source heterogeneous data improves the accuracy of disruptive technology topic identification. By integrating papers and patent heterogeneous data sources and combining the LDA topic model and multidimensional indicators (topic similarity, novelty and intensity), this study effectively identifies hot, emerging, potential and mature technology topics in the field of quantum communication. The empirical results show that the complementarity of multi-source data significantly reduces the limitations of a single data source (such as the delay in patent citations) and provides a more comprehensive scientific basis for the forward-looking identification of disruptive technologies.

• The classification of subject headings is helpful to identify the dynamic evolution characteristics of disruptive technologies. Based on the quantitative analysis of topic novelty and intensity, it is found that hot topics (such as quantum information carrier optimization) and highly novel emerging topics (such as machine learning assisting quantum communication) are more likely to breed disruptive technologies; mature topics (such as the basic research of quantum teleportation) show the characteristics of technical path solidification. This classification framework provides theoretical support for technology life cycle management and resource allocation.

• LDA topic model is helpful to the future industrial strategic layout. The LDA topic model and multi-dimensional index fusion method proposed in this study are not only applicable to the field of quantum communication but can also be extended to other cutting-edge technology fields (such as artificial intelligence and biotechnology).

While the classification framework effectively identifies emerging topics with high novelty but low current intensity, the critical question for both theory and practice is: How do these nascent themes evolve into full-fledged disruptive technologies? Understanding this transition mechanism is essential for moving beyond static identification toward dynamic monitoring and proactive management of disruptive potential. Drawing on the study’s findings and integrating insights from innovation theory, we propose a multi-phase mechanism through which emerging technology topics traverse the path from peripheral novelty to mainstream disruptiveness.

In this initial phase, the technology exists as a high-novelty, low-intensity topic, often confined to a small number of research groups or exploratory patents. Its characteristics align with the emerging topics identified in our framework (e.g. Topic WOS_0 on machine learning-assisted quantum communication). The transition mechanism at this stage is driven by three key processes:

1. Scientific legitimation. The technology must demonstrate scientific plausibility through rigorous peer-reviewed publications. Early papers establish the theoretical foundations and initial experimental validations, creating a credible knowledge base that attracts the attention of other researchers. This process of scientific legitimation is crucial for the topic to survive initial skepticism and gain a foothold in the academic community.

2. Community formation. A small but dedicated community of researchers begins to coalesce around the topic. This is evidenced by increasing co-authorship networks, specialized workshops or conference sessions and the emergence of shared terminology and research questions. This community serves as the “innovation nucleus,” driving incremental improvements and exploring diverse applications.

3. Artifact development. The transition from abstract concept to tangible artifact begins. Researchers develop prototypes, publish method papers and file initial patents. These artefacts, even if rudimentary, serve as proof-of-concept demonstrations that the technology can be realized in practice, moving beyond purely theoretical speculation.

As the community grows and initial artifacts are developed, the topic may transition into the potential topic category, characterized by focused research on specific applications and persistent challenges (e.g. Topic WOS_2 on QKD protocols). The mechanism driving this phase involves:

• Problem-solution pairing. Research becomes increasingly focused on solving specific, well-defined problems. For quantum communication, this includes developing protocols resistant to noise, improving key distribution rates or enhancing security against specific attack vectors. This problem-solution pairing narrows the scope of inquiry and directs effort toward demonstrable performance metrics.

• Niche market exploration. Although mainstream market adoption is distant, researchers and early-stage companies begin to identify and explore niche applications where the technology’s unique attributes provide immediate value, even if overall performance is inferior to incumbents. For example, early quantum communication systems found niche applications in securing critical infrastructure links where absolute security outweighed cost and speed considerations. This niche exploration is critical for generating early use cases and attracting initial, often government or defense-related, funding.

• Technology standardization efforts. Informal standards begin to emerge within the research community. Common protocols, benchmarking practices and shared experimental setups facilitate comparison and cumulative progress. These nascent standards reduce fragmentation and create a common platform for subsequent development.

Successful navigation of Phase II leads to the hot topic stage, where both novelty and intensity are high (e.g. Topic WOS_3/DII_3 on quantum information carriers and systems). This is the critical juncture where a technology transitions from a promising research area to a genuine competitive threat to incumbents. The key mechanisms are:

• Technological trajectory inflection. Through cumulative incremental improvements, the technology reaches a performance inflection point where it becomes competitive with established technologies on key metrics valued by the mainstream market. For quantum communication, this could mean achieving key distribution rates, transmission distances and system costs that approach or surpass those of classical encryption methods for specific applications.

• Ecosystem building. A broader innovation ecosystem begins to form around the technology. This includes specialized suppliers of components (e.g. single-photon detectors), system integrators, software developers and early adopters. The emergence of this ecosystem reduces the cost and complexity of adopting the technology, creating a self-reinforcing cycle of improvement and adoption.

• Institutional support and standardization. Formal standards are developed by industry consortia or standards bodies, reducing uncertainty for potential adopters. Government policy may shift from basic research funding to procurement programs, demonstration projects and regulatory changes that favor the new technology. This institutional support provides legitimacy and reduces market entry barriers.

In the final phase, the technology becomes a mature topic, characterized by high intensity but declining novelty. The disruptive outcome is realized through:

1. Mainstream market penetration. The technology successfully penetrates the mainstream market, displacing incumbent technologies in its primary application domains. This is driven by continued cost reductions, performance improvements and the accumulation of field experience that builds user confidence.

2. Architectural consolidation. A dominant design emerges, stabilizing the core architecture of the technology and shifting competitive focus to price, reliability and complementary services. This consolidation marks the transition from technological competition to market competition.

3. Spillover effects and new applications. The mature technology begins to generate spillover effects, enabling new applications and creating new markets that were previously impossible. For quantum communication, this could include integration with quantum computing networks for secure distributed computing, or the development of quantum sensing applications enabled by the same underlying infrastructure.

This mechanistic understanding of the transition from emerging to disruptive technologies has several important implications:

• Temporal dynamics of indicators. Our static classification based on novelty and intensity provides a snapshot of a technology’s position along this evolutionary path. By tracking how topics migrate across categories over time, we can create a dynamic monitoring system that provides early warning of accelerating trajectories.

• Differentiated intervention strategies. Each phase demands a different management and policy approach. In Phase I, the priority is funding diverse, exploratory research and fostering community formation. In Phase II, the focus should shift to supporting problem-solution pairing through targeted grants and facilitating niche market exploration through procurement programs. In Phase III, the emphasis moves to ecosystem building through standards development, infrastructure investment and market-creating policies.

• Identifying “False Dawns” and “Sleeping Giants”. The framework also helps distinguish between technologies that generate early excitement but fail to progress (emerging topics that never gain intensity) and those that remain under the radar for extended periods before sudden takeoff (potential topics with persistent but low-intensity research). Understanding the mechanisms that differentiate these trajectories is a critical direction for future research.

By embedding this mechanistic perspective within our classification framework, we transform it from a descriptive tool into a dynamic model of technological evolution, providing both researchers and practitioners with a deeper understanding of how today’s emerging topics can become tomorrow’s disruptive realities.

The primary contribution of this study lies not only in its methodological innovation but also in its capacity to generate actionable intelligence for a range of stakeholders. By translating the identified technology topics and their classification into specific decision-making contexts, this framework can serve as a practical tool for policymakers, corporate strategists and R&D managers.

The framework provides a systematic, data-driven foundation for crafting national science and technology policies, moving beyond reliance on expert intuition alone.

The classification of technology topics into hot, emerging, potential and mature categories offers a structured overview of a technological landscape. For instance, in the quantum communication case study, the identification of emerging topics (e.g. Topic WOS_0 on machine learning-assisted quantum communication) provides an early warning system for policymakers. This signal justifies the proactive allocation of dedicated, high-risk research funds through programs like “young scientist” grants or exploratory research schemes, nurturing potential breakthroughs before they become obvious to the broader community. Conversely, the identification of hot topics (e.g. Topic WOS_3 on quantum information carriers) validates and reinforces current funding priorities, ensuring that areas with high research intensity and novelty receive sustained support to maintain a competitive edge.

The framework enables a more dynamic and evidence-based approach to managing a nation’s technology portfolio. The clear demarcation of mature topics signals that the foundational research phase has been largely accomplished. Policymakers can use this information to strategically pivot funding from basic research towards applied development, proof-of-concept demonstrations and industry–university collaborations aimed at commercializing these mature technologies. This prevents the “technology lock-in” trap and ensures that resources are continuously cycled from fundamental discovery to economic and social impact.

The framework can be re-run at regular intervals to create a “living” map of the technological frontier. This allows policymakers to monitor how the landscape evolves in response to international developments, such as new regulations, foreign investments or geopolitical tensions. A sudden increase in the novelty or intensity of a topic in a rival nation could trigger a strategic reassessment and a rapid, targeted policy response, such as launching a national challenge or fostering a public–private partnership to accelerate domestic capabilities.

For firms operating in high-technology sectors, the framework provides a critical input for strategic decision-making, helping them navigate uncertainty and optimize their innovation pipelines.

The topic classification serves as a powerful input for corporate scenario planning. For example, a telecommunications firm exploring quantum communication technologies can use the framework to assess different investment scenarios:

If the firm’s analysis confirms that hot topics are central to its core business, it might decide on a strategy of aggressive internal investment to build a proprietary advantage.

If it identifies a high-novelty emerging topic that is promising but peripheral to its current capabilities, a more appropriate strategy might be corporate venture capital, investing in or partnering with startups that specialize in this area, thereby placing a strategic bet while managing risk.

For mature topics, the firm might focus on operational excellence, seeking to reduce costs, improve manufacturing processes and secure intellectual property through defensive patenting.

The framework can be operationalized as a systematic technology scouting tool. Corporate R&D teams can use the topic models to scan the global landscape, identifying not only which topics are hot, but also who the key players are (universities, startups, competitors) within each topic cluster based on their publication and patent outputs. This intelligence directly feeds into open innovation strategies, enabling targeted partnerships, licensing opportunities or the acquisition of promising startups. For instance, a firm interested in the “potential topic” of new quantum communication protocols could use the framework to identify the leading research groups in that niche for collaboration.

At the operational level, the framework provides R&D managers with a transparent and defensible logic for prioritizing projects and balancing their innovation portfolios.

The topic classification can be integrated into stage-gate project management processes. A project proposal aligned with an emerging topic might be evaluated primarily on its potential for high-impact novelty and its “option value,” with lower initial expectations for immediate market returns. Conversely, a project aligned with a hot topic would be judged on its potential to deliver a competitive advantage in a clearly defined and growing area. Projects aligned with mature topics would need to demonstrate a clear path to cost reduction, performance improvement or process innovation to be approved.

The framework offers a powerful visual tool for assessing the balance of an organization’s R&D portfolio. An R&D director could map the organization’s current projects onto the 2 × 2 matrix defined by novelty and intensity. This visual representation would immediately reveal any imbalances, such as an over-concentration in mature, low-novelty areas (risk of obsolescence) or an under-investment in promising emerging topics (missed opportunities). This insight enables deliberate and strategic portfolio rebalancing to align with the organization’s risk tolerance and long-term strategic goals.

The framework provides a quantitative basis for distributing resources across the portfolio. For example, an organization might decide to allocate 50% of its budget to hot topics to secure its current competitive position, 30% to emerging topics to build future options and 20% to potential topics to explore high-risk, high-reward unconventional paths. These allocations can be dynamically adjusted as the topics themselves evolve and are reclassified in subsequent analyses.

By translating the abstract categories of technology topics into concrete decision-making heuristics, this framework bridges the gap between quantitative technology foresight and the practical realities of strategic management in both the public and private sectors. It equips decision-makers not just with a map of the future, but with a compass and a set of tools to navigate it.

This study was based on an analysis of the existing published literature and patent documents. It did not involve human participants, animal subjects or any personal data collection. Therefore, ethical approval was not required for this study.