# Prior Art Search and Reranking for Generated Patent Text

arXiv: https://arxiv.org/abs/2009.09132 · HTML: https://mlsanity.com/paper/2009.09132/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 19 Sep 2020.

## Built on by (most cited first, 8 of 8)

- [Summarization, Simplification, and Generation: The Case of Patents](https://mlsanity.com/paper/2104.14860/) (2021)
- [Evaluating Generative Patent Language Models](https://mlsanity.com/paper/2206.14578/) (2022)
- [PatentEval: Understanding Errors in Patent Generation](https://mlsanity.com/paper/2406.06589/) (2024)
- [BERT based patent novelty search by training claims to their own description](https://mlsanity.com/paper/2103.01126/) (2021)
- [The Effectiveness of Bidirectional Generative Patent Language Models](https://mlsanity.com/paper/2211.09690/) (2022)
- [BERT based freedom to operate patent analysis](https://mlsanity.com/paper/2105.00817/) (2021)
- [PatentVision: A multimodal method for drafting patent applications](https://mlsanity.com/paper/2510.09762/) (2025)
- [Patentformer: A demonstration of AI-assisted automated patent drafting](https://mlsanity.com/paper/2510.09752/) (2025)

## Abstract (arXiv)

Generative models, such as GPT-2, have demonstrated impressive results recently. A fundamental question we'd like to address is: where did the generated text come from? This work is our initial effort toward answering the question by using prior art search. The purpose of the prior art search is to find the most similar prior text in the training data of GPT-2. We take a reranking approach and apply it to the patent domain. Specifically, we pre-train GPT-2 models from scratch by using the patent data from the USPTO. The input for the prior art search is the patent text generated by the GPT-2 model. We also pre-trained BERT models from scratch for converting patent text to embeddings. The steps of reranking are: (1) search the most similar text in the training data of GPT-2 by taking a bag-of-word ranking approach (BM25), (2) convert the search results in text format to BERT embeddings, and (3) provide the final result by ranking the BERT embeddings based on their similarities with the patent text generated by GPT-2. The experiments in this work show that such reranking is better than ranking with embeddings alone. However, our mixed results also indicate that calculating the semantic similarities among long text spans is still challenging. To our knowledge, this work is the first to implement a reranking system to identify retrospectively the most similar inputs to a GPT model based on its output.