STTR Phase I: A DLT Machine Learning Platform for Blockchain Warehousing

这一小型企业技术转让(STTR)项目的广泛影响/商业潜力改善了大型数据库、数据仓库和用于商业数据挖掘这些资源的软件的使用。在大数据时代，机器学习和人工智能等许多应用程序严重依赖于包括效率、互操作性和分析在内的数据功能。然而，由于集中式存储、同构数据格式和紧密耦合的工作流等多种技术限制，他们的数据子系统面临着满足这些需求的挑战。该STTR项目将开发一个新的框架来克服这些限制，以改善大数据在各个科学领域的应用，如生物科学、天文学和计算化学。在此第一阶段项目结束时，将产生开发基于区块链的数据仓库中间件所需的知识和基础材料。STTR第一阶段项目建议从现代数据仓库中发现的瓶颈开始，通过为DLT和区块链的进一步发展创造新的创新环境来推动DLT(分布式分类账技术)和区块链的具体知识和商业化。这里开发的DLT工具推动了区块链协议的发展，特别是在传统硬件上消除了基于正统协议的方法的二次和投机性成本，并使用其他创新的编程方法(例如，概率剪枝)建立了可线性化和恒定的成本。将为这些目的研究和开发新的统计、拓扑和计算方法，包括开发适用于机器学习、人工智能、Peta级和Exa级计算、高级科学计算以及未来几代教学学术发展的技术。预期结果包括一套新的协议和一个统一的工具，用于将数据类型和多个网络集成到传统数据仓库中，特别是对于努力跟上新的区块链数据、元数据和实时分析的数据仓库的进步。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

The broader impact/commercial potential of this Small Business Technology Transfer (STTR) project improves the use of large-scale databases, data warehouses, and the software that used to commercially data-mine these resources. In the era of big data, many applications, such as machine learning and artificial intelligence, critically rely on data functionalities including efficiency, interoperability, and analysis. However, their data subsystems are challenged to meet these needs due to multiple technical limitations, such as centralized storage, homogeneous data formats, and tightly coupled workflows. This STTR project will develop a new framework to overcome these limitations to improve big data applications in various scientific fields, such as biological sciences, astronomy, and computational chemistry. At the end of this Phase I project, the requisite knowledge and foundational materials for developing a blockchain-based data warehousing middleware will be produced.This STTR Phase I project proposes to advance the specific knowledge and commercialization of DLTs (Distributed Ledger Technologies) and blockchains by creating novel and innovative environments for further development for DLTs and blockchains as a "virtuous cycle", starting with bottlenecks identified in modern data warehouses. The DLT tool developed here advances blockchain protocols, specifically for removing quadratic and speculative costs from orthodox protocol-based approaches to blockchains on conventional hardware and instituting linearizable and constant costs with other innovative programming methods (e.g., probabilistic pruning). New statistical, topological, and computational approaches will be researched and developed for these purposes, including for development of techniques applicable for machine learning, artificial intelligence, peta-scale and exa-scale computing, advanced scientific computing, and pedagogical academic development of future generations. The expected results include a new set of protocols and a unified tool for integrating data types and multiple networks into conventional data warehouses, especially for advancement of data warehouses struggling to keep pace with new blockchain data, metadata, and real-time analytics.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.