SBIR Phase I: Agent-based Cognitive Assistant for Enabling Intelligent Space Operations

这个第一阶段项目的更广泛的影响/商业潜力是高效地处理极大的数据集。对于个人或团队来说，纯粹丰富的航空数据太过强大，无法通过记忆和经验来控制和控制。该项目将开发一个系统，通过将机器学习(ML)应用于不同类型数据集成的大数据集，提供额外的支持。它可以在系统开发的所有阶段帮助民用和军用做好准备、访问和速度，因为它降低了生命周期成本，并改善了生产、测试、验证、操作和支持期间的信息访问。此功能可应用于航空航天和非航空航天测试环境、长时间航天任务以及大型文档集中的信息需要即时访问并提供最佳可操作情报的其他环境。这个小型企业创新研究(SBIR)第一阶段项目演示了基于代理的认知辅助在真实世界空间操作环境中的应用，集成了机器学习(ML)和自然语言处理(NLP)分析工具。为了提高人类决策的速度和降低成本，可以利用搜索、索引、计算语言学以及广义地说机器学习和自然语言处理方面的最新发展来简化和改善大量高技术文档语料库的可访问性。该项目将展示，所有类型的大型数据集，包括多种格式的文本、数字的、地理空间的、结构化的和非结构化的，都可以使用先进的分布式开放源码搜索引擎进行查询。通过分析NLP ML算法产生的元数据，增强了搜索引擎结果的信息价值。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 Phase I project is to process extremely large data sets efficiently. The sheer abundance of aerospace data is simply too overwhelming for an individual or team to have command of and control from memory and experience. This project will develop a system that can provide additional support through machine learning (ML) applied to big data sets where different types of data are integrated. It can aid civilian and military readiness, access, and speed during all phases of system development, as it reduces lifecycle cost and improves information access during production, test, verification, operations, and support. This capability can be applied to aerospace and non-aerospace test environments, long spaceflight missions, and other contexts where information in a large document set requires instant accessibility and provides optimal actionable intelligence. This Small Business Innovation Research (SBIR) Phase I project demonstrates the application of agent-based cognitive assistance in a real-world space operations environment, integrating machine learning (ML) and Natural Language Processing (NLP) analytical tools. To improve the speed and lower the cost of human decision making, the latest state-of-the-art developments in search, indexing, computational linguistics, and broadly speaking, machine learning and natural language processing, can be leveraged to simplify and improve the accessibility of a huge corpus of highly technical documentation. This project will demonstrate that all types of large datasets, including text in a multitude of formats, numerical, geospatial, structured and unstructured, can be interrogated using an advanced distributed open-source search engine. The information value of the search engine results is enhanced by analyzing metadata produced from NLP ML algorithms.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.