A Lean 4 system models patent claims as DAGs with match scores in a verified complete lattice and supplies kernel-checked certificates for coverage calculations and five IP use cases, conditional on unverified ML inputs.
DAPFAM: A Domain-Aware Family-level Dataset to Benchmark Cross-Domain Patent Retrieval
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
citation-role summary
method 1
citation-polarity summary
years
2026 3verdicts
UNVERDICTED 3roles
method 1polarities
use method 1representative citing papers
Introduces a feature-level annotated patent dataset and LLM retrieval-reasoning workflows that outperform embedding baselines on passage retrieval and novel feature identification while avoiding spurious correlations in novelty prediction.
QaECTER sets new state-of-the-art patent retrieval performance on the new Sophia-Bench benchmark and an external test, outperforming a 23x larger general model and all prior patent-specific models using citation-driven training.
citing papers explorer
-
Formally Verified Patent Analysis via Dependent Type Theory: Machine-Checkable Certificates from a Hybrid AI + Lean 4 Pipeline
A Lean 4 system models patent claims as DAGs with match scores in a verified complete lattice and supplies kernel-checked certificates for coverage calculations and five IP use cases, conditional on unverified ML inputs.
-
Is It Novel and Why? Fine-Grained Patent Novelty Prediction Based on Passage Retrieval
Introduces a feature-level annotated patent dataset and LLM retrieval-reasoning workflows that outperform embedding baselines on passage retrieval and novel feature identification while avoiding spurious correlations in novelty prediction.
-
Citation-Driven Multi-View Training for Patent Embeddings: QaECTER and Sophia-Bench
QaECTER sets new state-of-the-art patent retrieval performance on the new Sophia-Bench benchmark and an external test, outperforming a 23x larger general model and all prior patent-specific models using citation-driven training.