Dunn, Alexander | Search NASA

DOE OSTI2024

Structured information extraction from scientific text with large language models

Extracting structured knowledge from scientific text remains a challenging task for machine learning models. Here, we present a simple approach to joint named entity recognition and relation extraction and demonstrate how pretrained large language models (GPT-3, Llama-2) can be fine-tuned to extract useful records of complex scientific knowledge. We test three representative tasks in materials chemistry: linking dopants and host materials, cataloging metal-organic frameworks, and general composition/phase/morphology/application information extraction. Records are extracted from single sentences or entire paragraphs, and the output can be returned as simple English sentences or a more structured format such as a list of JSON objects. This approach represents a simple, accessible, and highly flexible route to obtaining large databases of structured specialized scientific knowledge extracted from research papers.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗

NASA NTRS2019

MISSE-11 Ground Experiment Control

Back during the Apollo missions, when astronauts were being sent to the moon to explore and further scientific knowledge by conducting experiments and collecting samples for scientists back on Earth, they were faced with a surprising problem. It was discovered that the moon’s surface is covered with dust that has electrostatic properties, which would stick to suit and equipment. This hindered the functionality of the astronauts’ space suits, solar panels, optical instruments, among other exposed surfaces due to the jagged geometry of the dust which would damage said equipment. To resolve this issue, the Electrostatics and Surface Physics Laboratory (ESPL) came up with the solution of using an Electrodynamic Dust Shield (EDS)1 that uses electrostatic forces to move particles across a surface. Varieties of this dust shield have been developed for different applications. The purpose of the Materials International Space Station Experiments 11 (MISSE-11) is to experiment a payload that contains the dust mitigation technology to be flown to space for one year and an identical payload will be on earth, under vacuum, in the Electrostatics and Surface Physics Laboratory. The space payload was prepared prior to our arrival and shipped to fly on the International Space Station (ISS). The payload staying on the ground is going to act as a control for the experiment so that we may compare it to the payload that was flown in space. During our time at Kennedy Space Center, we were tasked with building and start testing the ground control unit of the MISSE-11 payload. To complete our task for the ground control unit, we constructed a new vacuum chamber setup and added automated aspects, procured the necessary flight electronics, designed and communicated for the machining of the payload frame, and performed functional and thermal testing on the electronics to the same original operational standards. Some modifications were made where possible, without affecting the performance of the samples, to be efficient with the production of the control unit while keeping important aspects identical. The experiment is now nearly complete, and testing will be starting within the next few weeks.

MISSE↗