Patents serve as a repository of technical and commercial knowledge, and protect intellectual property, playing an important role in promoting technological progress, business development, and innovation. These legal documents grant inventors temporary exclusive rights over their creations, incentivizing the disclosure of technical information that might otherwise remain hidden as trade secrets (Carr, 1995). By allowing inventors to profit from their creativity while simultaneously inspiring further technological advances through the revelation of prior art, patents directly impact both scientific and economic development (Hall, 2007; Langinier and Moschini, 2002; Schankerman, 1998).

Despite the rapid increase in academic literature exploring patents or leveraging patent documents and data, a current and comprehensive overview that captures the landscape of patent research holistically has yet to emerge. Such a panoramic perspective is invaluable for several reasons. First, it can reveal emerging topical clusters and current research trends, guiding scientists and practitioners toward areas of growing relevance. Second, mapping the landscape of patent research reveals the role and potential applications of more specialized subfields like patent analytics. This insight allows researchers developing quantitative patent analysis methods to focus their efforts on domains that stand to benefit most from such techniques. Finally, an academic landscape may facilitate cross-pollination across traditionally siloed disciplines by exposing potential applications of patent analytics.

To address this gap, our study aims to provide a landscape of patent research within academic literature. By surveying scholarly literature, we uncover major research topics, identify their interrelationships, and track evolving trends over time. We pay particular attention to the distinct and complementary studies of patent analytics, which have grown increasingly important for understanding innovation and economic progress. Specifically, this article addresses the following research questions:

In this article, we refer to “patent research” as any study that employs and leverages patents or patent data in any form and for any purpose, while “patent analytics” is used with a narrower scope to refer to studies with a more systematized approach to the study of patents or patent data, especially when used to develop metrics or methodologies (Daim et al., 2006). Patent analytics offers valuable insights into technology development trends, key industry players, and competitive landscapes through various approaches and techniques designed to extract meaningful information from patent data.

Our study employs a combination of data extraction techniques and topic analysis methods, including citation network analysis of scholarly articles. We present an overview of the current landscape, focusing on research fronts characterized by recency, relevance, and rapid growth (Rotolo et al., 2015). We expect to contribute by providing a comprehensive map of the patent research landscape to guide future studies and collaborations, identify emerging trends and underexplored areas to inform research priorities and funding decisions, provide insights into the evolving role of patent analytics to enhance evidence-based strategic planning and innovation policies, and develop a framework for integrating diverse patent research methodologies to foster interdisciplinary approaches in both academia and practitioners.

The remainder of this article is structured as follows: first, we explore the relevance of patents and scholarly patent research in general, while covering previous efforts in mapping the field of patent analytics. The methods section details our data extraction and analysis techniques. In the results section, we present our findings on the current landscape of patent research, with a focus on emerging trends and key areas of development. We conclude by discussing future directions for interdisciplinary research and shifts in methodological approaches within the field of patent research and analytics.

The academic interest in patent data spans several decades, evolving from early information retrieval systems to sophisticated analytical approaches. In the 1950s, pioneering work by Mooers (1952) laid the foundation for patent retrieval systems, initially focusing on searching metadata fields such as author, title, and keywords. As technology advanced, the scope expanded to include full-text analysis of patent documents in the 1960s and 1970s.

The 1970s and 1980s marked a significant shift in patent research, with scholars beginning to use patent statistics as a proxy for innovation and technological change. Soete (1979) and Pavitt (1985) examined the relationship between research and development (R&D) investment and patent counts at the national level, finding significant correlations. Other studies explored patenting patterns across countries and industries to understand differences in innovative activity (Evenson, 1984; Schiffel and Kitti, 1978; Sláma, 1981). At the firm level, Pakes and Griliches (1980) conducted one of the first systematic analyses of the relationship between R&D and patenting, finding a strong cross-sectional relationship but weaker time-series correlations.

A landmark contribution came from Trajtenberg (1990), who studied the computed tomography scanner industry. By combining patent data with market information, Trajtenberg demonstrated that while raw patent counts correlated poorly with social value creation, citation-weighted patent counts showed a strong correlation (around 0.75) with total social welfare created. This finding has been corroborated by subsequent studies, such as Harhoff et al. (1999) and Hall et al. (2005), establishing the importance of patent citations as indicators of economic and technological significance.

The analysis of patent citation networks emerged as a distinct field of study in the latter half of the 20th century. Early work by de Solla Price (1965) highlighted the importance of citation analysis in understanding scientific and technological development. The 1980s and 1990s saw the formalization of quantitative approaches, with Narin (1994) introducing various patent metrics for the study of Innovation. The release of the NBER Patent Citations Data File in 1990 provided researchers with a comprehensive dataset, spurring further studies on knowledge spillovers and innovation diffusion (Hall et al., 2001).

More recently, patent analytics has expanded its applications across various domains of technology management and innovation policy. Key areas include competitive intelligence, technology forecasting, R&D planning, merger and acquisition analysis, and policy evaluation. The exponential growth in global patent data, with 2022 alone estimated at 3.46 million patent applications worldwide (WIPO, 2023), has needed the development of more sophisticated and automated methods for analysis. Thus, the field has benefited from the integration of advanced techniques such as text mining, natural language processing, network analysis, and machine learning. This plurality of methodologies and scopes has led to the emergence of various terms describing the field, like patent bibliometrics (Narin, 1994), patinformatics (Trippe, 2003), and technology mining or tech mining (Porter, 2004), each with nuanced scopes and target applications, reflecting its multidisciplinary nature.

The use of patents by academics has been surveyed in the past, with the work of Basberg (1987) being an early example. This survey focused on the use of patents to measure technological change. Scholars at the time were concerned with the use of patent citations, finding “important” patents, and benchmarking innovation across regions. A comprehensive survey by Griliches (1990) reviewed several decades of research on patent statistics as economic indicators. He examined multiple data sources, including patent counts, renewal data, and stock market valuations. His survey synthesized evidence from studies using the U.S. Patent Office data, European patent renewal information, and firm-level R&D expenditure data highlighting critical measurement challenges, including the highly skewed distribution of patent values and variations in patenting propensity across industries and time. Griliches' synthesis helped establish methodological frameworks for evaluating patent quality and understanding the limitations of patent statistics as innovation indicators. More recent efforts have adopted computer-assisted methods to bring a more systematized understanding of the field by using bibliometrics (Mejia et al., 2021). Mikova (2016) analyzed Global TechMining conference proceedings from 2011 to 2015, identifying trends such as the integration of multiple approaches (e.g., bibliometrics, NLP, statistical analysis) and the use of novel data sources (e.g., web data, social media). Aristodemou and Tietze (2018) reviewed 57 articles on applying AI, machine learning, and deep learning to intellectual property data, categorizing them into knowledge management, technology management, economic value, and information extraction/management. The study found a growing interest in intellectual property (IP) analytics but called for more research on use cases and firm-level applications. Karata et al. (2024) analyzed 1,006 papers on “patent analysis,” revealing trough a descriptive approach that “technology” was the most common keyword and that top journals included “Technological Forecasting and Social Change” and “Information Processing & Management.” Hu et al. (2024) explored the foundations and frontiers of technology mining using co-citation analysis, bibliographic coupling, and content analysis of 277 articles. The study identified text analysis, bibliometrics, patent analysis, and strategic technology management as foundational areas, with technology topic analysis, roadmapping, component analysis, opportunity analysis, and management/decision support as frontier clusters.

While these previous studies offer valuable insights, they are limited in providing a comprehensive understanding of patent research given their narrower econometric or methodological focus. Our study aims to address these gaps by analyzing a larger and more recent dataset to capture the latest developments. It takes a holistic view of patent research, considering all aspects rather than focusing on specific methods or applications.

The bibliographic data for this study was sourced from the Web of Science (WOS) Core Collection. To identify relevant articles, a topical search was conducted using the query TS = “patent^*”, where the asterisk serves as a truncation symbol to accommodate variations of the term (e.g., patents). The search was performed without time constraints, retrieving articles from all available years in the database. Data were retrieved on May 31, 2024, yielding 103,738 articles.

While comprehensive, the query also retrieves articles unrelated to the target topic due to the various meanings of the term “patent”. In addition to referring to intellectual property documents, “patent” may be used as an adjective to denote open, unobstructed, or accessible, particularly in biomedical research. For example, there is extensive research on patent ductus arteriosus, a congenital heart defect (Schneider and Moore, 2006). Other deviating meanings include its use as a synonym for obvious, clear, or apparent. From a document retrieval standpoint, it may be tempting to generate a list of banned keywords (e.g., to be used with the NOT operator), but this would result in neglecting patent analytics papers on those alternative meanings [for instance, patent analysis of patent ductus research (Hsieh et al., 2004)]. Therefore, to focus on our target topic, citation networks were employed as both a data-cleaning mechanism and a means to extract thematic clusters.

Academic articles are positioned within a research field by citing previous related research. Articles that do not cite nor are cited by other articles were excluded from the study, as these are the papers that used the keyword “patent” without belonging to the patent research domain. A direct citation network was constructed, establishing linkages between articles when one cites the other (de Solla Price, 1965). Direct citation networks are known to surface research field taxonomies (Klavans and Boyack, 2017) and help identify research fronts (Shibata et al., 2008), making them suitable for long-term bibliometric research. However, this network would also contain papers in other fields of research, such as in biomedicine, that may cover other meanings of patents. To exclude these, thematic clusters were extracted, and after human inspection, unrelated clusters were pruned from the citation network.

Identifying topics from a citation network involves grouping nodes with denser connections compared to other groups. An optimal partition is achieved when the link density is higher at the intra-cluster level than the inter-cluster level, maximizing the network's modularity (Newman, 2006). We applied the Louvain method, a computationally efficient algorithm for partitioning large networks, to obtain the clusters (Blondel et al., 2008). For large networks, the first pass of the clustering algorithm may result in relatively large clusters. To obtain a more granular view, we applied the resolution limit theorem (Fortunato and Barthélemy, 2007) to further split clusters into subclusters, resulting in a topical hierarchy that facilitates the analysis, as the topics become smaller and more coherent. The authors named the clusters based on an assessment of the titles of the most connected articles, the most frequent keywords, and relevant metadata such as journal names, countries, or authors. During this step, the final cleaning was conducted, and unrelated clusters were removed from the study. Figure 1 represents a summary of the methodology.

Summary statistics of the publication years and citations received by the articles within each cluster and subcluster were calculated. Concretely, we apply metrics related to “size” being the number of documents; “relevance” the standardized cluster citations; “emergence” the average publication year; and “fast growth” the largest delta increase of publication counts within each subcluster over the past 10 years. These metrics are known to be useful in defining emerging research (Rotolo et al., 2015). Subclusters that are outliers in any of those metrics are separated for detailed descriptions.

Bibliographic data, including the full record and cited references, were exported as tab-delimited files from the WOS website. The dataset was processed using the statistical software R version 3.6.3 (R Core Team, 2019), with the igraph package version 1.2.5 (Csárdi et al., 2024) for network creation and clustering and the tm package version 0.7.7 for text processing (Feinerer et al., 2008). The citation network was visualized using the large graph layout (Adai et al., 2004), selected for its computational efficiency. The choice of layout has no impact on the research results.

From the original dataset, 53,668 articles used alternative meanings of the term “patent” that are not part of the core of patent research. These papers are disconnected from the main corpus of knowledge as they do not cite or receive citations from other patent-related literature. The rest of the articles compose the citation network of patents-related literature, covering 50,070 articles. Figure 2 shows the citation network forming a 2-group divide of clusters, one being those from biomedicine fields where patent is used in its medical meaning, and the other group of clusters studying patents as the IP document. Patent as the IP document has the largest share with 27,119 (54%) articles. In Figure 2, this group looks smaller due to its more cohesive nature, as the citations seem to be shared across the clusters in the group. In the remaining of this article, we refer as the citation network of patents research to this group of 27,119 documents and the results presented hereafter are based on this dataset.

The study of patents in academic literature has been steadily growing over the past decade, with an average 7% yearly increase in publications. The earliest records in the dataset of the divide date back to 1909, when Baekeland (1909) called for revisions to the United States (US) patent law. Since then, the trend of publications has been on the rise, with 2,139 publications in 2023 alone. The US and China have been the largest contributors to patent-related literature, with China consolidating its position as the country with the most publications since 2020. In 2023, more than 35% of publications came from China, while the US followed at 17%. The United Kingdom (UK) has consistently maintained its position as the third-largest contributor.

The study of patents spans across various fields, with Economics, Management, and Law being the most prominent. The journals that have been at the forefront of publishing patent-related literature include Research Policy, Reviews on Therapeutic Patents, Scientometrics, Technological Forecasting and Social Change, and Sustainability. Supporting figures for these descriptive statistics are presented in the Supplementary material.

Through a citation network analysis of over 27,000 academic articles, 15 main clusters and 93 subclusters were identified, revealing the landscape of patent research. The results show a steady increase in the use of patent documents in academic research. We note that the interest in academia is shared by two distinct but highly integrated groups, one that focuses on management and innovation studies and the other that is more applied to pharma and biomedical research. For instance, clusters such as “Patent Analytics and Innovation Dynamics” and “Advanced Methods in Patent Analytics and Technology Forecasting” dominate the first, while “Patent Systems and Biomedical Innovations” dominates the second. Emerging trends in environmental sustainability and biomedical innovations were identified, as evidenced by the recent and rapidly growing subclusters in these areas. The analysis also revealed the widespread integration of advanced analytical techniques, including AI and machine learning, across various domains of patent research. These findings provide a foundation for addressing the study's core research questions: how patent documents are used in academic research, current trends in patent research, and the role of patent analytics methods within the broader scope of patent research. The following discussion examines these questions in detail, drawing insights from the identified clusters and their interrelationships.

How is patent information used in academic research?

We found that patent bibliographic data and information are used extensively and diversely in academic research, serving as rich sources of information for understanding technological innovation, knowledge flows, and economic impacts. The cluster analysis revealed several key applications of patent data in research:

Indicators of innovation: Patents are widely used as proxies for measuring innovative activity across different technological fields and geographic regions. This being the most mature use of patents in Innovation Management, as evidenced by the seminal work of Soete (1979) and later consolidated by Narin (1994) highlighting the similarities between patent bibliometrics and scientific literature bibliometrics, particularly in assessing national and inventor productivity.

Knowledge spillover analysis: The network visualization of subclusters (Figure 6) highlighted the importance of patent citations in tracing knowledge flows as inferred from the central place of subclusters 1-2 and 6-1, both on patent citation flows. These build upon the seminal work of Jaffe et al. (1993), who used patent citations to demonstrate the geographic localization of knowledge spillovers. Our findings suggest that this approach remains relevant, with studies on geographical proximity and spillover still being conducted and what has changed is the sophistication of the methods or the access to new datasets.

Technological forecasting: Several subclusters focus on using patent data to identify emerging technologies and predict future trends. Some of these, like most of the subclusters in cluster 3, prime the development of methods from patent data, while other subclusters focus on the application of such methods to bring forward-looking views on the fields (e.g., research on patent analytics for energy, green energy, and nano technologies) This streams aligns with the tech mining approaches (Porter and Newman, 2009), now being extended by the incorporation of more advanced data analytics techniques.

Innovation quality assessment: This covers research using patents as a benchmark to assess firms or innovation portfolios. Although there is a prominent background in patent counts and citations, some research emphasizes the multidimensional nature of patent quality, challenging simplistic metrics, thus aligning more with the concept of patent quality (Higham et al., 2021). Our analysis supports this view, showing diverse approaches to evaluating patent impact across different technological domains (e.g., as in subclusters 1-1, 8-4).

Policy and economic analysis: Patents are used to assess the effectiveness of R&D funding strategies. We observed subclusters focused on how different funding sources (public vs. private) influence patenting outcomes across various sectors (Shelton and Leydesdorff, 2012) and also research areas focused on the interplay between industry, academia, and government trough different initiatives of polices (Ivanova et al., 2017).

Corporate strategy and competitive intelligence: Patent analysis is extensively used to understand competitive landscapes and inform strategic decision-making. Most of the clusters related to innovation management are geared toward this use case.

Interdisciplinary research: We observe the integration of patent analysis with diverse fields such as environmental sustainability, biomedical innovations, and digital technologies, showcasing the versatility of patent data in being a supporting data source for multiple disciplines.

What are the current trends in patent research?

Current trends in patent research reflect a shift toward more sophisticated, context-aware approaches in patent analysis. This evolution addresses Meyer's (2000) call for recognizing the unique characteristics of patent citations compared to scientific citations. The emergence of tech mining and AI-driven analytics in patent research, as highlighted in recent and growing subclusters, suggests a shift toward more data-intensive and nuanced approaches to extracting value from patent documents.

A prominent trend is the integration of environmental sustainability and green innovation with patent analytics. Subclusters focusing on “Patent Analytics in Carbon Reduction Technologies” (9-3) and “Drivers of Corporate Environmental Innovation” (9-4) have shown recent emergence and significant growth. This trend reflects the broader societal focus on sustainable development and demonstrates how patent analysis is being applied to track and foster eco-friendly innovations.

The rise of micro-level scientometrics, focusing on detailed interactions within organizations and among individuals, represents another current trend. This approach, highlighted in Zhang et al. (2017) work on scientometrics for tech mining, is evident in growing subclusters that examine the impact of corporate structures, funding sources, and individual inventor characteristics on innovation outcomes.

Another notable trend is the application of advanced data analytics and artificial intelligence in patent research. Subcluster 3-4, “Patent Analytics in Energy Sectors,” exemplifies this trend, showcasing the use of sophisticated analytical techniques to understand technological evolution in energy-related fields. While this subcluster is more on the applied side, it is embedded within cluster 3 of advanced methods, suggesting a need in the field of energy innovation to leverage the most up-to-date methodologies for innovation analysis. The integration of patent analytics into unrelated but specific research has been long foreseen as part of the discussions on tech mining (Porter and Newman, 2009).

Lastly, we observe a growing trend in analyzing the intersection of patent data with other data sources, such as scientific publications, market data, and policy information. This holistic approach to innovation analysis, evident in subclusters like 1-5 “Strategic Alliance Governance and Innovation Outcomes,” represents an evolution from earlier, more siloed approaches to integrated patent analysis. Advances in natural language processing facilitate the integration of diverse data sources.

Role of patent analytics methods within the larger scope of patent research?

The role of patent analytics methods within the larger scope of patent research has evolved to become increasingly central and sophisticated. Patent analytics methods serve as critical tools for identifying and quantifying technological emergence (Carley et al., 2018). Growing subclusters focused on advanced methods in patent analytics indicate a shift toward more data-driven and objective approaches to understanding innovation trajectories.

These methods facilitate the integration of diverse data sources, allowing for a more comprehensive understanding of innovation ecosystems. There is a trend in patent analytics methods now routinely combining patent data with scientific publications (Mejia and Kajikawa, 2020), market information, and even social media (Orduna-Malea and Font-Julian, 2022) data to provide richer insights.

Patent analytics methods play a crucial role in enhancing the strategic value of patent research for both policymakers and industry practitioners. The growth of subclusters focused on competitive intelligence and strategic decision-making underscores how these methods are bridging the gap between academic research and practical applications.

This study provided an overview of the current state and future directions of patent analysis in academic research. We identified 93 research streams from academic literature that use the patent document in any form; these topics were evaluated in terms of size, recency, citation impact, and growth, revealing relevant trends. These include an increased focus on AI methods and the application of patent analytics for sustainability and evaluation of corporate performance. We further organized the topics to reach a five-core component framework encompassing fundamentals of patent systems, patents as indicators, methodological developments, IP management practices, and applications. By proposing an integrated framework and identifying key trends and challenges, we contribute to both the theoretical understanding and practical application of patent analytics.

As patent data becomes increasingly accessible and analytical techniques more sophisticated, the field of patent analysis is poised to play an even more crucial role in informing innovation policy, guiding corporate strategy, and advancing our understanding of technological progress.

The evolution of patent analysis from simple citation counts to complex, AI-driven analyses reflects the growing recognition of patents as rich sources of technological and economic information. However, this evolution also brings new challenges in terms of data interpretation and methodological rigor. Future research should focus on addressing these challenges while continuing to explore novel applications of patent analytics across various domains of science and technology and also in various sectors, such as academia, business, and policy, to empower innovation.