Hyperspherical Semantic Trajectory Analysis: Mapping Technological Diffusion across Academic Preprints, Patent Signals, and Compute Scaling
Researchers propose HSTA to track technology diffusion across 30,000 arXiv preprints and USPTO patents.
The paper introduces Hyperspherical Semantic Trajectory Analysis (HSTA), which projects Transformer sentence embeddings onto unit hyperspheres using Spherical K-Means and UMAP. It studies 30,000 filtered arXiv preprints and USPTO patent records and defines Semantic Centroid Vector Drift and Commercialization Offset. Large Language Models show the highest drift at 0.332, ahead of Artificial Intelligence Systems at 0.234. Vector autoregression tests find quarterly paper-volume velocity alone does not Granger-cause frontier compute surges in Epoch AI data at conventional significance.
- Analyzes 30,000 filtered arXiv preprints and USPTO patent records
- LLM sub-topic semantic drift is 0.332, the highest of eight topics
- Paper-volume velocity alone does not Granger-cause compute surges
- Ties textual signals to Epoch AI hardware and compute metrics
From the original article
1 min read at huggingface.co
Macroeconomic productivity metrics, such as Total Factor Productivity, register technological breakthroughs with multi-year reporting lags due to administrative survey intervals and national accounting conventions.
By Muhammad Sukri Bin Ramli · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.