PATENTWRITER: A Benchmarking Study for Patent Drafting with LLMs
This work addresses the tedious patent-filing process for legal and innovation professionals, but it is incremental as it applies existing LLMs to a new domain-specific task.
The paper tackles the problem of automating patent abstract generation by evaluating six leading LLMs, including GPT-4 and LLaMA-3, under various prompting strategies, and finds that they can produce high-fidelity and stylistically appropriate abstracts, often outperforming domain-specific baselines.
Large language models (LLMs) have emerged as transformative approaches in several important fields. This paper aims for a paradigm shift for patent writing by leveraging LLMs to overcome the tedious patent-filing process. In this work, we present PATENTWRITER, the first unified benchmarking framework for evaluating LLMs in patent abstract generation. Given the first claim of a patent, we evaluate six leading LLMs -- including GPT-4 and LLaMA-3 -- under a consistent setup spanning zero-shot, few-shot, and chain-of-thought prompting strategies to generate the abstract of the patent. Our benchmark PATENTWRITER goes beyond surface-level evaluation: we systematically assess the output quality using a comprehensive suite of metrics -- standard NLP measures (e.g., BLEU, ROUGE, BERTScore), robustness under three types of input perturbations, and applicability in two downstream patent classification and retrieval tasks. We also conduct stylistic analysis to assess length, readability, and tone. Experimental results show that modern LLMs can generate high-fidelity and stylistically appropriate patent abstracts, often surpassing domain-specific baselines. Our code and dataset are open-sourced to support reproducibility and future research.