Reproduce “From Rules to Neural Graphs: Scalable Structured Prediction for Patent Prior Art Search” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Nikolai Zenovkin, Sebastian Björkqvist
The authors introduce a neural parser that converts patent text into invention graphs for prior-art retrieval, using local word-pair scoring to scale from short-text training to long documents without retraining. In the Graph Transformer system, reparsing queries and a 3-million-document corpus subset raises citation-averaged Recall@3 from 0.3336 to 0.3373; full-corpus scaling remains in progress.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.