The knowledge graph is an infrastructure originally constructed by google in 2012 to enhance the google search engine, by semantically linking information gathered from a wide variety of data sources.…

This is a grant award from the National Aeronautics and Space Administration (NASA) to Southern Methodist University (SMU) to design and develop a prototype NASA Earth Science Knowledge Graph. The objective is to advance NASA's knowledge search and discovery capabilities by semantically linking different distributed information objects, such as data, journal articles, services, tools, and people, within the NASA Earth Science Enterprise. The project aims to leverage approaches like information extraction, machine learning, and graph databases to uncover novel relationships and insights from NASA's Earth science data and content. The award has a potential value of $186,797.00 and does not have a set-aside designation.

The knowledge graph is an infrastructure originally constructed by google in 2012 to enhance the google search engine, by semantically linking information gathered from a wide variety of data sources. The concept of the knowledge graph has since evolved as a strategy by which to uncover information and knowledge buried within the vast amounts of text data on the internet. There emerged novel relationships and information that can be surfaced using information extraction and inference methods, machine learning techniques such as WORD2VEC (mikolov, 2013), graph databases, and so on. The knowledge graph approach can be utilized to semantically link information gathered from a wide variety of sources within the NASA earth science enterprise for earth science concepts. Applications of knowledge graph methods will allow for contextually richer knowledge extraction, enhanced keyword and metadata predictions, deep edge traversals and relationship predictions. Objectives inspired by google s knowledge graph, this project aims to design and develop a prototype NASA earth science knowledge graph to connect all the distributed disparate information objects in NASA s earth science enterprise. Our primary goal is to advance the state-of-the-art NASA knowledge search and discovery capability by going beyond the traditional catalog search for data, and semantically linking the different distributed information objects. These information objects range from data, journal articles, services, tools and people. We have already begun developing the prototype that extracts and links information from journal publications and NASA STI content about hurricane science.