Efficient Patent Searching Using Graph Transformers

TL;DR

This paper introduces a Graph Transformer-based dense retrieval method for patent searching that models inventions as graphs, improving accuracy and efficiency over traditional text embedding models by leveraging domain-specific relationships and examiner citations.

Contribution

The paper presents a novel graph-based retrieval model that enhances patent search accuracy and efficiency by incorporating invention graphs and examiner citation data.

Findings

Significant improvement in prior art retrieval accuracy.

Enhanced computational efficiency for long document processing.

Outperforms existing text embedding models in patent search tasks.

Abstract

Finding relevant prior art is crucial when deciding whether to file a new patent application or invalidate an existing patent. However, searching for prior art is challenging due to the large number of patent documents and the need for nuanced comparisons to determine novelty. An accurate search engine is therefore invaluable for speeding up the process. We present a Graph Transformer-based dense retrieval method for patent searching where each invention is represented by a graph describing its features and their relationships. Our model processes these invention graphs and is trained using prior art citations from patent office examiners as relevance signals. Using graphs as input significantly improves the computational efficiency of processing long documents, while leveraging examiner citations allows the model to learn domain-specific similarities beyond simple text-based matching. The result is a search engine that emulates how professional patent examiners identify relevant documents. We compare our approach against publicly available text embedding models and show substantial improvements in both prior art retrieval quality and computational efficiency.