1. Learn
  2. Articles
  3. The IMPACT of Summer Interns

The IMPACT of Summer Interns

Summer interns supporting NASA’s Interagency Implementation and Advanced Concepts Team (IMPACT) helped advance several IMPACT machine learning initiatives.

Derek Koehl, IMPACT Science Writer

Each year NASA’s Interagency Implementation and Advanced Concepts Team (IMPACT) benefits from the expertise and energy of summer interns. This summer’s group of interns – Bethany Kuo, Anisha Kabir, and Tony Park – are an excellent example of the contributions interns make to project teams over the course of 10 weeks and the valuable career-related skills they gain from working closely with experienced mentors and scientists in the fields of Earth science and computer science.

Image of multi-colored connected dots showing relationships between the connected elements.

Example of a knowledge graph IMPACT intern Bethany Kuo generated from a single Earth science research paper. NASA IMPACT image.

IMPACT is a component of NASA’s Earth Science Data Systems (ESDS) Program and works to expand the use of NASA Earth observing data through innovation, partnerships, and technology. This summer’s interns applied their knowledge and insight to projects in IMPACT’s machine learning group. One of this group’s focus areas is research into how to apply natural language processing (NLP) and knowledge graph technologies to improve the discovery of knowledge within Earth sciences. These three interns fit naturally into this effort.

Bethany Kuo is a computer science and linguistics double major at the University of Maryland. In collaboration with IMPACT team members, she explored the effectiveness of combining a knowledge graph with bidirectional encoder representations from transformers (BERT) sentence embeddings and graph convolutions to surface related information to provide answers to natural language queries.

“I am interested in learning more about natural language processing, where the two fields of computer science and linguistics intersect,” Kuo says. “This project has the potential of making it easier to find information about Earth science phenomena along with the related papers.”

Side-by-side image of line graphs on left and word cloud examples on the right.

Left panel: Graphs of nine top topics in AGU ESSI abstracts and how they have grown over time. Right panel: Word clouds of these nine topics with their most frequent keywords. NASA IMPACT image.

During her time with IMPACT, Anisha Kabir, a rising junior studying computer science at the University of California, Santa Barbara, used machine learning to identify and analyze trends of key topics in American Geophysical Union (AGU) Earth and space science paper abstracts over the past 15 years. As she explains, this entailed web scraping AGU Earth and Space Science Informatics (ESSI) abstracts from 2005 to 2020 along with their author and index term information. Using the latent topic modelling method, she clustered the abstracts into 50 different topic groups and created trend plots for further analysis. The goal of the project is to provide better insight into key topics and trends in the Earth and space science informatics community over time.

Image of Anisha Kabir.

IMPACT intern Anisha Kabir.

“I got interested in machine learning and natural language processing through my university courses and research, but I’ve wanted to learn how these can be applied to different scientific fields,” says Kabir. “This internship was an incredible opportunity for me to see the intersection of Earth science and machine learning, and I’ve learned a lot about both fields.”

Tony Park, who is a rising junior pursuing a computer science degree at the University of Michigan, worked on an IMPACT project that seeks to extract specific information (science topic, machine learning algorithm, problem type) from Earth science journal articles. He contributed to the creation of a named entity recognition (NER) model for the labelling process. This involved manually labelling hundreds of abstracts to create a training dataset and then using that dataset as input to improve the model. He also assisted in coding the model using Python and used pretrained models and algorithms provided by Google.

Headshot of Tony Park

IMPACT intern Tony Park.

“I’ve always been fascinated by the idea of robots doing tasks on their own, and I wanted to take my interest a step further by researching the potential of machine learning and adapting to the environment around it,” Park says. “Ever since I graduated high school, I have been focused on artificial intelligence and machine learning, and I am grateful to have this opportunity for an internship at NASA.”

IMPACT Manager Dr. Rahul Ramachandran commented on the accomplishments and contributions of these emerging young scientists.

“I have been impressed by how quickly each of the interns integrated themselves into the collaborative environment here at IMPACT and learned our work practices,” he says. “Without question, each intern has meaningfully contributed to our mission of maximizing the effectiveness and efficiency of Earth science data management and stewardship.”

These three interns demonstrate why IMPACT remains committed to investing in and fostering the next generation of scientific talent. All of us here at IMPACT are grateful for having had the opportunity to work with them.


Information about NASA's internship program and an electronic application can be found on NASA's Internships and Fellowships website. More information about IMPACT can be found on NASA's Earthdata website and the IMPACT project website.

Published August 12, 2021

Page Last Updated: Aug 12, 2021 at 10:00 AM EDT