Identifying technological trajectories using machine learning algorithms

该项目旨在提高我们对科学知识为什么在某些方向上发展的理解，以及是什么导致它在各个领域增长得更快或更慢。该项目将创建新的神经网络机器学习算法，以扫描专利和科学文章文本（说明书和权利要求书），寻找自然语言概念。这些结果可能会被专利局用来改进他们自己的数据库搜索，被商业分析公司用来揭示新技术和潜在的合作者，被学术经济学家用来了解知识是如何传播和积累的。

This project aims to improve our understanding of why scientific knowledge progresses in certain directions and what causes it to grow faster or slower across fields. The project will create new neural-network machine-learning algorithms to scan patent and scientific article texts (specifications and claims) for natural language concepts. The results will potentially be used by patent offices to improve their own database search, by business analytics companies to reveal new technologies and potential collaborators, and by academic economists to understand how knowledge travels and accumulates.