Source-linked AI summary
Interactive Overlay Maps for US Patent (USPTO) Data Based on International Patent Classifications (IPC)
Loet Leydesdorff, Duncan Kushnir, Ismael Rafols
TL;DR
Patent analysis needs ways to connect hierarchically classified technologies with citation-based relationships across large datasets. The paper develops public-domain USPTO basemaps and flexible overlays at 3-digit and 4-digit IPC levels, enabling comparisons across patent sets and time. The resulting maps support distinct high-level and fine-grained analyses, while USPTO download limits constrain data retrieval.
Problem
Hierarchical IPC co-classifications can be difficult to aggregate and may be sensitive to class changes, motivating citation-based mapping of USPTO technologies.
Method
The study builds 3-digit and 4-digit USPTO citation basemaps, cosine-normalizes citation matrices, and provides tools for overlaying searched patent sets.
Results
The 3-digit maps support high-level analysis, whereas 4-digit maps provide finer resolution for specific technology analysis; overlays can also be animated across years or domains.
Takeaways & Limitations
The interface enables analysis of technological distance, portfolio diversity, and developments in technologies or organizational capabilities over time.
Takeaways & Limitations
USPTO limits downloads to 1,000 patents per run, requiring subsequent downloads and careful file handling for larger searches.
Abstract
from arXiv · showhide
We report on the development of an interface to the US Patent and Trademark Office (USPTO) that allows for the mapping of patent portfolios as overlays to basemaps constructed from citation relations among all patents contained in this database during the period 1976-2011. Both the interface and the data are in the public domain; the freeware programs VOSViewer and/or Pajek can be used for the visualization. These basemaps and overlays can be generated at both the 3-digit and 4-digit levels of the International Patent Classifications (IPC) of the World Intellectual Property Organization (WIPO). The basemaps can provide a stable mental framework for analysts to follow developments over searches for different years, which can be animated. The full flexibility of the advanced search engines of USPTO are available for generating sets of patents and/or patent applications which can thus be visualized and compared. This instrument allows for addressing questions about technological distance, diversity in portfolios, and animating the developments of both technologies and technological capacities of organizations over time.
1. Introduction
The paper frames patent mapping as a way to connect technologically organized patent data with citation-based overlays, while addressing limits of hierarchical co-classification maps. It develops USPTO basemaps organized by IPC citation relations for flexible analysis of technological distance and portfolios.
- Patent data and classification: Patents are organized by manifest classification systems, unlike scholarly literature, which is commonly organized through journals and disciplines.IPC classifications provide an explicit technological organization for patent data.
- Related overlay approaches: Earlier overlay approaches positioned document sets within disciplinary or journal-category maps, while related work extended overlays to medical, geographic, and patent data.These studies motivate applying overlay techniques to technological categories.
- Research problem: The study maps USPTO data through citations among IPC classes rather than WIPO co-classifications, whose hierarchical structure can make aggregated co-classification maps difficult to interpret.The authors also note that class splitting can introduce uncontrolled indexer effects into co-classifications.
- Research problem: Citation networks vary in density within and across clusters, whereas IPC classifications impose a hierarchy; this makes citation networks potentially less sensitive to misclassification than co-classification networks.The distinction reflects the network’s additional degree of freedom relative to a hierarchy.
- Study contribution: The paper provides public-domain USPTO basemaps at 3-digit and 4-digit IPC levels using approximately 4.2 million patents and 39 million citation relations from 1976-2011.VOSViewer and Pajek are supplied for visualization, with flexibility for further user adjustments.
2. Methods and materials
The study constructs 3-digit and 4-digit IPC citation basemaps from USPTO patents and supplies tools for generating overlays. Citation distributions are cosine-normalized, mapped visually, and used to calculate technological distance and diversity.
- Data preparation: Primary IPC classes and citations are organized into asymmetric citing-versus-cited matrices at both 3-digit and 4-digit levels.The resulting matrices support separate basemaps at the two classification resolutions.
- Normalization: The citation matrices are cosine-normalized using the citing side because citing is treated as the active variable, while the cited dimension represents the complete archive.Cosine normalization is useful for sparse matrices because it is insensitive to their many zero values.
- Visualization: VOSViewer supplies clustering and visualization, while Pajek files and other network-analysis programs extend the available mapping options.VOSViewer avoids the threshold required by earlier spring-embedded mappings because it uses an MDS-like solution.
- Mapping and distance: Technological distance is defined as dij = (1 – cosineij), and the mapping program projects citation-vector space into two dimensions by minimizing a stress function.The projected map represents distances among IPC categories while preserving the multivariate structure as closely as possible.
- Diversity: Rao-Stirling diversity is computed by summing technological distances weighted by the proportions of the respective categories.The routine writes diversity values locally for both IPC levels.
3. Results
The paper provides 3-digit and 4-digit IPC basemaps from normalized citation relations and uses them to visualize patent and patent-application sets as overlays. The interface supports comparative, animated analysis across searches and years, while VOSViewer and Pajek provide alternative visualization options.
- Basemaps: The basemaps cover 124 IPC classes at three digits and 630 classes at four digits, with the latter also represented as a density map.The 3-digit basemap uses approximately 39 million citations among 4.2 million US patents from 1976–2011.
- Basemaps: Three-digit labels support high-level analysis, whereas four-digit labels may be needed for specific technology analysis.The two map resolutions exhibit the same structural dimensions despite being flipped horizontally and vertically, as MDS solutions may be rotated and translated.
- Overlay maps: USPTO advanced searches can combine inventor, address, applicant, title, abstract, and date fields with AND and OR operators to generate overlay sets.The routines organize downloaded patents or applications into VOSViewer input files at both IPC levels, with proportional counting of IPC attributions.
- Overlay maps: Applications provide a view closer to the research front because patent examination and granting can take several years.The application routine parses downloads from the applications interface and produces maps corresponding to patent overlays.
- Overlay maps: The 2007 Dutch-address patent and application distributions correlate at r = .753 (p < .01), while their Rao-Stirling diversity is almost similar.The corresponding four-digit correlation is r = .648 (p < .01); stable basemaps enable animation across years and domains by changing node sizes and prominent labels.
4. Alternatives
The paper compares patent-mapping programs and extends its USPTO interface with Pajek-compatible outputs, community detection, and thresholded IPC basemaps.
- Program comparison: The interface accepts flexible USPTO searches and organizes downloaded patent sets for visualization in VOSViewer and Pajek.Searches must retrieve more than fifty patents before the dedicated programs automate subsequent processing.
- Program comparison: Alternative programs differ in visualization and network-statistics capabilities; Gephi and VOSViewer provide superior visualization, while Pajek and Gephi support additional network analysis.The comparison motivated compatibility with Pajek and, through Pajek, Gephi.
- Pajek extensions: Pajek-compatible outputs add community detection and other network statistics unavailable in VOSViewer.The routines also provide vector and cluster files for visualizing patent contributions and labeling communities.
- Pajek extensions: With cosine > 0.2, the largest components contain 109 of 124 three-digit classes in 11 communities and 605 of 630 four-digit classes in 21 communities.The corresponding modularity values are Q = 0.529 and Q = 0.681; users can alter thresholds, partitioning, and coloring.
- Pajek extensions: Figure 6 shows the same patent set in Pajek using Kamada & Kawai layout, Blondel coloring, and exclusive labeling for one community.The example contains 1,908 US patents with an inventor in the Netherlands, published in 2007, at the three-digit level.
5. Conclusion
The paper presents IPC-based USPTO basemaps and flexible overlays as instruments for studying technological development across firms, nations, and technologies over time. It frames these tools as part of a broader research program connecting databases and tracking trajectories dynamically.
- Conclusion: The authors provide USPTO basemaps and overlay tools at three-digit and four-digit IPC levels for following developments across databases.The tools support both coarse-grained and fine-grained technological mapping.
- Conclusion: The instruments address a data-silo problem by supporting transversal translations between institutional and cognitive contexts.The paper identifies cross-connections between longitudinally organized datasets as important for studying innovation.
- Conclusion: IPC organizes patents intellectually while balancing technological refinement against user-friendliness and providing a baseline across major patent systems.The paper describes USPTO fields as enabling mapping and animation for patent analysis.
- Conclusion: The tools support quantitative and animated studies of technological competencies, diffusion, diversification, and trajectories across science, technology, and innovation.The authors specifically mention measures such as Rao-Stirling diversity and animations of development processes.
- Conclusion: Compared with earlier aggregated and static OECD analyses, the instruments enable detailed dynamic study of individual firms, nations, and emerging technologies.The paper gives changing IPC positions of nanotechnology patents as a possible follow-up application.