Executive Summary
This method introduces a systematic approach to identify and score patents at risk of suppression by analyzing multi-dimensional metadata patterns including filing timelines, classification anomalies, citation networks, and legal event histories. It enables users to prioritize patents for deeper investigation based on quantifiable suppression risk scores.
Problem
Existing methods to identify suppressed patents rely heavily on anecdotal evidence or manual, inconsistent analysis of patent documents and legal events. There is no standardized, testable approach to quantify suppression risk using patent metadata.
Hypothesis (Evidence-first)
Suppressed patents exhibit measurable metadata patterns distinct from non-suppressed patents, such as unusual classification changes, extended prosecution timelines, or citation anomalies, which can be quantified to produce a suppression risk score.
What’s New (Non-duplicate Angle)
- Combines multiple metadata dimensions into a unified scoring system for suppression risk.
- Uses legal event timelines and classification changes as key indicators.
- Incorporates citation network anomalies to detect potential suppression.
- Provides a testable, repeatable scoring methodology rather than qualitative judgment.
Apparatus / Setup
- Patent database access — Access to comprehensive patent metadata including filing dates, classification codes, legal event histories, and citation data
Notes: Preferably from USPTO, EPO, or WIPO bulk data - Data analysis software — Statistical or data science tools such as Python with pandas, R, or specialized patent analytics platforms
Notes: Required to process and analyze large metadata datasets - Scoring algorithm implementation — Custom scripts or software implementing suppression risk scoring based on defined heuristics
Notes: Modular design to allow tuning and validation - Validation dataset — Set of patents labeled as suppressed or non-suppressed from literature or expert annotation
Notes: Needed to test scoring accuracy
Procedure (BETA)
- Collect a representative sample of patents including known suppressed and regular patents.
- Extract metadata for each patent: filing dates, prosecution timelines, classification code changes, citations, and legal event histories.
- Define heuristics for suppression indicators, e.g., prolonged prosecution, frequent classification reassignments, low citation counts despite technical relevance, unusual legal event patterns.
- Implement a scoring algorithm that assigns weighted scores to each heuristic for every patent.
- Calculate suppression risk scores for the dataset.
- Validate scoring by comparing with known suppressed patents and adjust weights to improve discrimination.
- Document scoring thresholds that optimize true positive suppression detection.
- Apply the scoring method to new patents to prioritize candidates for further manual analysis.
Measurement Plan (Success Criteria)
- Accuracy of suppression risk scoring: Compare predicted suppression risk against expert-labeled suppressed/non-suppressed patents using precision, recall, and F1-score.
Tools: Statistical analysis software; confusion matrix computation
Success: F1-score above 0.7 indicating useful discrimination - Repeatability of scoring: Re-run scoring on same dataset and verify consistent scores.
Tools: Version-controlled scripts and datasets
Success: Zero variance in scores across repeated runs - Coverage of metadata features: Percentage of patents with complete metadata enabling scoring.
Tools: Data completeness reports
Success: At least 90% data completeness for scoring - Usability of scoring output: User feedback on clarity and actionable value of risk scores for patent analysis.
Tools: Surveys or structured interviews with practitioners
Success: Majority (>75%) positive feedback on usefulness
Risks & Safety Boundaries
- No physical risks; data privacy must be respected when handling proprietary patent information.
- Ensure compliance with database licensing and usage terms.
- Avoid drawing legal conclusions solely from risk scores; scores are indicators for further analysis.
- Responsible disclosure if sensitive suppressed patent data is uncovered.
- No encouragement of patent infringement or illegal use.
Legal & Ethics
- Respect intellectual property laws and patent confidentiality.
- Use data only from publicly available or licensed databases.
- Avoid defamatory claims about specific patents or holders based on scoring results.
- Disclose that suppression risk scoring is a heuristic and not definitive proof of suppression.
Sources to Verify (Evidence Pack)
- (No external sources captured in this run.)
Claims that Require Verification
- Claim: Suppressed patents exhibit distinct metadata patterns that can be quantified.
Check: Empirical studies or expert analyses supporting metadata patterns in suppressed patents.
Sources: https://patentlyo.com/, https://papers.ssrn.com/, WIPO Publications
BETA Protocol (A/B)
- A Variant: Use classification change patterns as primary indicator with fixed weights.
- B Variant: Use citation network anomalies as primary indicator with fixed weights.
- Duration: 30 days
- Participant Notes: Participants should have access to patent metadata and basic data science tools. Feedback on scoring usability and accuracy is essential.
Report Template
Suppression Risk Scoring Report: dataset description, heuristic definitions, scoring results, validation metrics, user feedback summary.
Vote
Which metadata dimension do you consider most indicative of patent suppression?
- Prolonged prosecution timelines
- Frequent classification changes
- Citation network anomalies
- Unusual legal event patterns
0
voters
Next Iterations
- Integrate machine learning classification models to improve scoring accuracy.
- Expand dataset to include international patents and cross-jurisdiction analysis.
- Develop visualization tools for suppression risk mapping.
- Incorporate expert feedback loops for continuous scoring refinement.
Tier target: Practitioner | BETA: True