Tag
The paper introduces Hyperspherical Semantic Trajectory Analysis (HSTA), an unsupervised methodology for tracking technological diffusion using semantic embeddings from academic preprints and patent documents.
This paper proposes ClaimGAT, a Graph Attention Network that encodes patent claim dependencies to predict litigation risk, achieving an AUC-ROC of 0.818 on USPTO utility patents.
The article describes a DIY project where the author built a Moire pattern-based device to guide ships, inspired by a patent and video, and discusses its practicality compared to simpler navigation systems.
An Isomorphic Labs researcher shared their experience using Gemini 4 to summarize papers on protein targets and break down 500-page chemistry patents, reducing tasks that originally took one to two days or even a month to just over ten minutes and a single afternoon, with extremely high accuracy.