Whether you're a patent attorney preparing for prosecution, an inventor assessing patentability, or a defendant building an invalidity case, knowing how to conduct a thorough prior art search is essential. A well-executed search supports novelty, obviousness, and freedom-to-operate analyses in line with guidance from authorities such as the USPTO and WIPO.
This step-by-step guide walks through a structured, defensible approach to prior art searching.
What Is Prior Art?
Prior art is any evidence that an invention was publicly known before a patent application’s priority date. As recognized by the USPTO and WIPO, it may include:
- Patents and patent applications (worldwide)
- Academic papers and journals
- Products sold or publicly used
- Technical standards and specifications
- Conference presentations with published proceedings
- Websites and online publications (with verifiable public dates)
The goal of a prior art search is to identify existing disclosures that either anticipate an invention or render it obvious when combined with other known references.
Step 1: Define Your Search Scope
Understand the Invention
Before searching, clarify:
- What problem the invention solves
- Its key technical features
- The relevant industry or technology domain
- Essential versus optional elements
Set Your Search Objectives
Different objectives require different depth and coverage:
| Search Type | Purpose | Depth Required |
|---|---|---|
| Novelty Search | Assess newness | High |
| Freedom to Operate | Identify infringement risks | Medium |
| Invalidity Search | Challenge granted patents | Very high |
| Landscape Analysis | Understand competitive space | Medium |
Step 2: Extract Keywords and Concepts
Analyze the Technology
Break the invention into searchable concepts.
Example: AI-powered patent search tool
- Core technology: artificial intelligence, machine learning, NLP
- Application: patent search, prior art analysis, IP research
- Methods: semantic search, vector embeddings, transformer models
- Outputs: relevance ranking, claim mapping
Create Keyword Lists
Technical Terms
- Primary: “artificial intelligence patent search”
- Synonyms: “AI patent analysis,” “automated patent search”
- Related: “semantic patent search,” “patent analytics”
Industry Terms
- “prior art search,” “patentability analysis”
- “freedom to operate,” “claim mapping”
Account for Terminology Differences
Terminology varies across:
- Academic literature (formal terms indexed in Google Scholar)
- Patents (broad, abstract language)
- Industry publications
- Standards documents
Step 3: Choose Your Databases
Patent Databases
Free
Professional
- PatSnap
- Derwent Innovation
- Orbit Intelligence
- Traindex for analytics-driven patent and technology insights
Non-Patent Literature Databases
Academic
- Google Scholar
- IEEE Xplore
- PubMed
- arXiv
- Semantic Scholar
Standards
- IEEE Standards Association
- IETF RFC Archive
- ISO Online Browsing Platform
Product and Commercial Sources
- Wayback Machine
- Company websites and documentation
- GitHub repositories (with commit dates)
Step 4: Execute Your Search Strategy
Start Broad, Then Narrow
Exploratory Phase
- Use 2–3 core keywords
- Review initial results
- Capture new terminology
Targeted Phase
- Add classification codes (CPC/IPC)
- Search by inventors and assignees
Exhaustive Phase
- Include foreign-language patents
- Expand into non-patent literature
- Review standards and archived products
Boolean and Advanced Searching
Step 5: Analyze and Document Results
Evaluate Relevance
- High: Direct anticipation
- Moderate: Partial disclosure
- Background: Contextual
Maintain a Search Log
Documenting search steps aligns with best practices discussed in Scopus-indexed patent research literature.
Map References to Claims
Create claim charts mapping invention elements to prior art disclosures.
Step 6: Expand the Search
- Follow backward and forward citations using Google Scholar or The Lens
- Review inventor publication histories
- Explore adjacent technologies
Step 7: Handle Special Cases
Software and AI
Search academic papers and open-source projects alongside patents.
Business Methods
Post-Alice v. CLS Bank, focus on technical implementation details rather than abstract concepts.
Step 8: Validate and Verify
Confirm:
- Priority dates
- Public accessibility
- Accurate translations for foreign-language references
Step 9: Use AI-Powered Tools
AI-driven platforms support semantic discovery and large-scale analysis:
- PatentScan for combined patent and non-patent literature search with claim mapping
- Traindex for patent analytics and technology landscape analysis
- The Lens for citation analysis
Step 10: Report Your Findings
A strong report includes:
- Executive summary
- Search strategy and scope
- Results analysis
- Claim-level comparisons
Conclusion
Effective prior art searching blends structured methodology with broad source coverage and careful documentation. While patents remain a core component, non-patent literature—academic papers, standards, and archived disclosures—often provides decisive evidence.
Platforms like PatentScan and Traindex help bridge patent databases and non-patent literature, enabling more complete and defensible prior art analysis while supporting the rigor required for prosecution, opposition, and litigation.
References
USPTO – MPEP §2128 (Printed Publications)
https://www.uspto.gov/web/offices/pac/mpep/s2128.htmlWIPO – Prior Art and Patentability
https://www.wipo.int/patents/en/topics/prior_art.htmlGoogle Scholar
https://scholar.google.comThe Lens
https://www.lens.orgScopus
https://www.scopus.comInternet Archive – Wayback Machine
https://web.archive.org
Top comments (0)