The Air Force Office of Scientific Research (AFOSR) has awarded a $450,000 grant to Assistant Professor of Computer Science Sandeep Kuttal. She is one of 36 scientists and engineers from 27 research institutions and businesses who submitted over 215 winning research proposals through the Air Force’s Young Investigator Research Program (YIP).
“On behalf of the College of Engineering and Natural Sciences (ENS), I heartily congratulate Professor Kuttal for receiving a grant through the Air Force’s YIP,” said Jim Sorem, the dean of ENS. “Since joining our faculty, Professor Kuttal has distinguished herself as an exemplary researcher and an outstanding mentor for our students. Her interdisciplinary investigations of the intersection between software engineering and human-computer interactions are fascinating and will, I have no doubt, break new ground in this swiftly evolving field.”
Kuttal’s research program is Supporting Information Foraging by Utilizing Agents’ Collective Foraging Behavior. The issue at the core of her work is how to optimize the sorting and analysis of a vast array of “information fragments” in order to produce “optimal information” that is meaningful and applicable to a specific goal.
In order to illustrate her quest, Kuttal points to the Department of Defense. There, intelligence officials are tasked with gathering information from a variety of sources (e.g., the internet, agents’ reports, federated data collections) to create “actionable intelligence.” The huge challenge, Kuttal explained, “is that analysts must pour through these collections and manually piece these fragments together. The inflow of information this creates can quickly overwhelm even the most experienced analyst.”
Seeking and foraging
For her project, Kuttal and her team will investigate the use of the past collective information-seeking behaviors of knowledge workers in order to reduce the burden of finding relevant information for newcomers working on similar tasks. To guide their inquiry, they will deploy information foraging theory. Previously, Kuttal, along with various collaborators, introduced Information foraging theory to two new domains: end-user developers debugging for visual web-based programs and developers foraging in the presence of information variants, which results in variations foraging theory.
“In modern times, mass communication, mass media and networking technologies have enabled collective intelligence of software development to span huge groups of developers, distributed across continents and time zones,” Kuttal explained. “The result has been the emergence of a global brain of knowledge that is capable of achieving great feats. These groups of developers collaborate together to develop complex projects with a common goal.”
In order to achieve such a goal, members of these groups often seek information from numerous different sources, such as the web, software artifacts and agents. This abundant information often makes the development of projects easier but, paradoxically, can make it more difficult due to the substantial costs of time and cognitive effort while information seeking.
Information seeking becomes expensive when there is too much information (“information overload”), the information is intermingled with potentially irrelevant and redundant knowledge (“information pollution”) or related information is scattered across different sources. “Developers need to manually sift through and piece together relevant information,” Kuttal commented. “Therefore, collective intelligence applications inherently face challenges when seeking optimal information.
Ants, bees and human developers
At the center of their project, Kuttal and her team will develop a system grounded in information foraging theory. For inspiration, they looked at collective intelligence patterns among non-human animals. According to Kuttal, most animals in the wild have developed ways of communicating information between individuals/groups when performing optimal foraging tasks. For example, ants use pheromone trails and stridulation, while honeybees use the celebrated waggle dance to recruit others to a food source.
“Similar to this behavior,” Kuttal noted, “we believe that developers also leave trails of their intelligence (e.g., commit logs) and their foraging behavior among artifacts (e.g., within open-source software and on the web) that can be captured and utilized. We aim to utilize the synergistic information associated with individual developers’ foraging interactions to generate a global intelligent-foraging data model that represents optimal information sources within software development platforms, such as open-source software, and in unpredictable environments, such as the web.”
Current models of this theory, Kuttal explained, account only for individual foragers’ behavior in isolation. “We intend, however, to extend the scope to include the costs and values of different artifacts to different foragers working on different tasks. From there, we’ll create data models that capture the collective mental models of groups of individuals. This will enable us to develop an unsupervised prediction algorithm based on these mental models and to use these predictions to make recommendations to specific individuals based on their personal foraging histories.” The end result should be an improvement in groups’ “collective foraging wisdom” that, in turn, will lead to better foraging recommendations.
Thanks to a Faculty Development Summer Fellowship from TU, Kuttal was able to conduct preliminary research during summer 2020. “I am thankful, also, for the constant support and feedback of my colleagues in the Tandy School of Computer Science,” she remarked. “Their insights and collegiality have been instrumental in furthering my research and career goals.”
Training opportunities
Kuttal’s YIP-supported research will also enable her to expand her research lab in order to provide opportunities for students – undergraduate and graduate – to address a number of problems associated with human-computer interaction (HCI), software engineering (SE) and artificial intelligence.
“Working independently and collaboratively, they will tackle open-ended problems, conduct simulations, analyze data and develop effective research-communication skills at key HCI and SE conferences,” noted Kuttal. She also underscored that among her proudest career accomplishments is having published scholarly papers with 11 undergraduate and 2 graduate students at TU during her five years at the institution.
The YIP is open to U.S. citizens and/or permanent residents who are scientists and engineers at U.S. research institutions who received a PhD or equivalent degrees in the last seven years and who show exceptional ability and promise for conducting basic research of military interests. The program’s objective is to foster creative basic research in science and engineering, enhance early career development of outstanding young investigators and increase opportunities for the young investigators to recognize the Air Force mission and the related challenges in science and engineering.
Distributed systems, cloud computing, software engineering, gaming and simulation, bioinformatics: these are just a handful of the areas awaiting your discovery at the Tandy School of Computer Science.