Key takeaways
• A low-result search does not establish novelty. Test search coverage before drawing a conclusion.
• Expand each core technical concept across terminology, functional wording, classifications, and document-based routes.
• Use a known relevant document to test retrieval, then record the search scope and remaining limitations.
Treat zero results as a coverage question first
Imagine that you search for a technical feature and find almost nothing relevant. It is tempting to conclude that the field is empty and the invention must be novel.
That conclusion skips an important question:
Did the search cover the ways patent documents are likely to describe the technology?
A patent search retrieves documents through the language, classifications, citations, sources, jurisdictions, and filters included in the strategy. Narrow routes can produce a narrow result set even when relevant documents exist.
Low result volume is not a substitute for coverage testing.
Why familiar technical language can miss patent literature
Patent documents may describe the same technical subject in different ways:
- Vocabulary and abstraction: a product-specific term may appear in a patent as a broader or narrower technical concept.
- Functional and structural wording: a document may describe what a component does, how it is arranged, or what result it produces instead of using the expected component name.
- Time, language, and translation: terminology changes as a field develops, and equivalent concepts do not always map to one literal translation.
Illustrative example
An inventor might describe a feature as:
heat dissipation structure for phones
A patent document might instead use:
thermal management assembly for portable electronic devices
The phrases do not show that the disclosures are the same. They show how closely related technology can use very different language. A one-phrase search may never retrieve the second document.
Eureka connects technical inputs with source-linked evidence for novelty and FTO review. Turn an invention or product description into structured, source-linked patent evidence for review.
Build a search that can learn from the literature
Start with the small number of concepts that make the invention work. Separate them from benefits, marketing language, and implementation details that would narrow the search too early.
1. Define the core technical concepts
Start with the small number of concepts that make the invention work. Separate them from benefits, marketing language, and implementation details that would narrow the search too early.
For a cooling phone case, the starting concepts might be:
- a case or housing for a portable device;
- heat transfer, cooling, or thermal management; and
- the particular mechanism that creates the technical distinction, if it is sufficiently defined.
The objective at this stage is to create a search model that can be expanded and tested, not to decide patentability.
2. Expand each concept along several language routes
For every core concept, consider:
- direct synonyms;
- broader and narrower terms;
- functional expressions;
- component, material, and process terms;
- historical terminology; and
- relevant language or translation variants.
Keep the relationships visible. A list of disconnected keywords is difficult to debug. A concept-to-term map shows which technical idea each term is intended to retrieve.
3. Add classification and document routes
Keywords should not be the only path into the literature. Relevant classifications can surface documents that use unexpected wording. Citations, patent families, applicants, inventors, and known competitors can provide additional routes when they are relevant to the task.
Classifications do not guarantee completeness. They can be broad, updated, or applied differently across offices. Their value is that they test the field through a structure other than literal wording.
4. Test the strategy with a known document
Select a document already known to be relevant to the technical subject and ask whether the current strategy retrieves it. This is a known-document coverage test, sometimes called reverse validation.
If the document does not appear:
- inspect the terminology used in its title, abstract, claims, and description;
- review its classifications and family members;
- identify which concept or route was missing; and
- revise the strategy before relying on its coverage.
Retrieving one known document does not prove that the field has been covered. Failing to retrieve it is direct evidence that the current strategy has a gap.
5. Iterate and document the limits
Each useful result can teach the next query. Record the terms, classifications, sources, jurisdictions, date limits, and exclusion decisions used in the search. Also record what was not searched.
That documentation lets another reviewer understand how the result set was produced and where additional work may be needed.
What a low result count can and cannot support
| A low result count may indicate | A low result count does not prove |
|---|---|
| The query is narrow | The invention is novel |
| The terminology is uncommon | No relevant prior art exists |
| The classification route needs refinement | An examiner will not find another document |
| The technology is emerging or specialized | The invention involves an inventive step |
| The selected sources or jurisdictions are limited | The search is complete |
Novelty remains a legal assessment of the claimed invention against the applicable prior art. Under the European Patent Convention, an invention is new if it does not form part of the state of the art. EPO guidance explains that a novelty objection requires a prior disclosure that directly and unambiguously makes the claimed subject matter available. A search result set supplies evidence for that assessment. It does not replace it.
From a search strategy to reviewable evidence
The current Eureka Novelty Search workflow begins with a natural-language invention description. It extracts technical features, builds multiple search strategies, and compares relevant prior art against those features. The supporting documents remain available for review.
The workflow can reduce the effort required to build and inspect multiple routes. Reviewers should still:
- check the extracted features and search strategies;
- test whether known documents can be recovered;
- open the cited source documents; and
- treat the output as evidence for professional judgment, not as proof that no other prior art exists.
Eureka connects technical inputs with source-linked evidence for novelty and FTO review. Turn an invention or product description into structured, source-linked patent evidence for review.
