Abstract

A large proportion of software maintenance effort is dedicated to understanding existing code. Program comprehension time could be reduced by helping developers skim code quickly and decrease the amount of code they need to read thoroughly. When programmers write internal comments, they serve as an abstraction of the high-level algorithmic steps that comprise the method and provide a quick overview of the high level actions described in an abstract way. Unfortunately, internal comments are rare and sometimes stale.  We have defined the concept of an action unit as a code block that consists of a sequence of consecutive statements that logically implement a high-level action, or high-level algorithmic step. A method typically consists of multiple action units, where an action unit is too small to be a single method, but requires more than one statement to implement. Beyond internal comments, another targeted application of the resulting high level descriptions of high-level algorithmic steps in a method is to improve method summaries.  This talk provides an overview of our work to automatically identify and generate natural language descriptions for specific types of high level algorithmic steps, by using structural and linguistic information embedded in the source code. 

Bio

Dr. Lori Pollock is Alumni Distinguished Professor in Computer and Information Sciences at the University of Delaware and ACM Distinguished Scientist. Her research focuses on software artifact analyses for easing software maintenance, testing, and developing energy-efficient software, code optimization, and computer science education. She serves as Cochair of the Computing Research Association’s Committee on Education (CRA-E) and on the Executive Board of the Computing Research of Women in Computing (CRA-W), which was honored with the National Science Board's 2005 Public Service Award to an organization for increasing the public understanding of science or engineering. She was awarded the ACM SIGSOFT Influential Educator award 2016 and University of Delaware's Excellence in Teaching Award, E.A. Trabant Award for Women's Equity in 2004.  She co-leads the Partner4CS team in Delaware to integrate CS into K-12 through teacher professional development and undergraduate service to teachers in the CS10K national efforts.