UC Berkeley students are using data science to make the U.S. Air Force more efficient

Introduced by Andre​ ​Almeida​ ​Reggiani,​ ​Neel​ ​Davar,​ ​Rivers​ ​Jenkins,​ ​Phillippe​ ​Phanivong,​ ​Jia​ ​Xi​ ​Tay 

Supervisors in the U.S. Air Force typically spend hours writing performance reports for each one of their employees. But they don’t have to anymore.

A group of five UC Berkeley students have been working with the U.S. Air Force to speed up the performance report writing process using natural language processing. The students’ new program, ComGen, converts plain English accounts of employee performance into properly formatted comments for U.S. Air Force reports.

Without ComGen, supervisors have to manually look-up and replace abbreviations, acronyms and synonyms to ensure each comment they write takes up one full line in their performance report. But with the new program, supervisors can generate several different ways to word their comment and choose the one that works best for their report.