FlowPlan-G2P: A Structured Generation Framework for Transforming Scientific Papers into Patent Descriptions

TL;DR

FlowPlan-G2P is a structured framework that transforms scientific papers into patent descriptions by decomposing the task into concept extraction, section planning, and graph-conditioned generation, improving legal compliance and quality.

Contribution

It introduces a hierarchical, graph-based generation approach specifically designed for converting scientific papers into legally compliant patent descriptions, outperforming existing models.

Findings

FlowPlan-G2P outperforms proprietary models on domain-specific benchmarks.

Structured decomposition improves legal compliance over scale alone.

Standard metrics favor non-compliant outputs, highlighting the need for domain-specific evaluation.

Abstract

Generating patent descriptions from scientific papers is challenging due to fundamental rhetorical and structural disparities between the two genres. Existing approaches treat this as surface-level rewriting, failing to capture the hierarchical reasoning and statutory constraints inherent in patent drafting. We propose FlowPlan-G2P, a graph-mediated generation framework that decomposes this transformation into three stages: (1) Concept Graph Induction, extracting technical entities and functional dependencies into a directed graph; (2) Section-level Planning, partitioning the graph into coherent subgraphs aligned with canonical patent sections; and (3) Graph-Conditioned Generation, synthesizing legally compliant paragraphs conditioned on section-specific subgraphs. Experiments on expert-validated benchmarks reveal that standard NLG metrics systematically favor legally non-compliant outputs over valid patent descriptions, motivating our domain-specific evaluation. Under this evaluation, FlowPlan-G2P with an open-weight backbone consistently outperforms vanilla proprietary models, demonstrating that structured decomposition is a stronger determinant of quality than model scale.