Dane Payba was born and raised in Maui and graduated from Maui High School as Valedictorian in 2020. He is currently going into his second year at the University of California, Los Angeles, majoring in Computer Science. Over the years, Dane has developed a deep passion for the Machine Learning and Artificial Intelligence field, developing numerous related projects, joining multiple competitions, and even an active part of the Association of Computing Machinery, Advanced AI pathway at his school. In his free time, he enjoys helping others, going to the beach, exploring the island, and playing tennis, even being a part of the club team at UCLA.

Home Island: Maui

High School: Maui High School, HI

Institution when accepted: University of California, Los Angeles

Akamai Project: Visualizing Natural Language Processing Results on Performance Work Statements

Project Site: KBR – Kihei, Maui

Mentor: Oliver Grillmeyer

Project Abstract:

Performance Work Statements (PWS) provide information for the work that is to be accomplished for specific contracts in the DoD. There exists a large corpora of historical textual data containing PWS that are ingested into digital format for advanced analytics. There is currently an HPC-backed machine learning solutions project deployed, utilizing Natural Language Processing (NLP) and BERT transformers to develop a big data analytics pipeline geared towards text analysis. At this state, this analysis is displayed solely in numerical form. The output is complex to digest for the user and would greatly benefit from a visual presentation of information. Visual results are much easier to process in contrast to the previously mentioned quantitative data and provide much greater insights into the structure of data. Word frequencies, combinations, and patterns are all examples of the current output. Utilizing the visualization software of Tableau and MatPlotLib, numerical results were processed to create charts, word clouds, distributions, graphs, and histograms based on the resulting text analysis after undergoing NLP. By obtaining visualizations of this data, meaningful analysis of the NLP model and PWS will be achieved. Examples include the ability to see most common words in the corpora of text, provide insights into the structure of textual data, obtain quick information without searching through numerical output, and the transformation of data through heatmaps, precision-recall curves, and confusion matrices. Through visual analysis, feedback about the machine learning model analysis can be enhanced, allowing the opportunity to better see the results of NLP processing.