EPA Salt March Project

Collaborators: Scott Jackson, Kate Ficas-Naleway, Amanda Davis, Ryan Wicks, Josh Ward, Lucy Cousins, Brett Barnard (all at UMass Amherst). 

One of the research questions we ask: Is it possible to detect salt marsh dieback using aerial imagery and machine learning?

In this project we utilize unoccupied aerial systems (UAS) or drones to acquire cross-seasonal multi-spectral imagery with the goal of developing fine spatial scale landcover classification of ten New England salt marshes. It is extremely challenging to map salt marsh vegetation, because of tidal dynamics. Traditional remote sensing (e.g., satellite imagery) doesn’t work because one cannot control the timing of image acquisition and the cell size is coarse (e.g., 10 square meters). UAS imagery overcome these issues. Over the last three field seasons, our team has systematically collected a large spatial-temporal multispectral image dataset for each marsh and have developed computational methods on the Holyoke High Performance Computing Cluster to analyze this data. We are now producing our first results: fine-scale landcover classifications of these marshes. These results could be a game-changer for the way scientists and policymakers monitor the effects of climate change and work to protect these vital coastal systems.

National Science Foundation Growing Convergence Research:
Jumpstarting Successful Open-Source Software Projects With Evidence-Based Rules and Structures

Collaborators: Brenda Bushouse (UMass Amherst), Vladimir Filkov (UC Davis), Seth Frey (UC Davis) 

Overarching research questions: How are open source software projects sustained? What leads some projects to successful “graduation” in OSS nonprofit foundation incubation programs while others do not?

It is estimated that more than 80% of the world’s computational processing is dependent on Open source software (OSS) – software that is not proprietary and is not sold for a profit. Given that these are not traditional firm-owned projects (think Microsoft or Apple) but the quintessential instance of CBPP, the key question is: How are OSS projects sustained? In fall 2020 our interdisciplinary team was awarded a five-year interdisciplinary “convergence” grant from the US NSF to discern the socio-technical structural and governance conditions under that sustain open source software projects working under a nonprofit organization’s umbrella. This project will advance our knowledge of how OSS projects are supported by overarching nonprofit organizations, but I expect the findings will have important and more broad insights into how CBPP can be sustained in the areas outside of software, including collaborative writing, open source hardware and open source science.

 

National Science Foundation Research Coordination Network:
Coordinating and Advancing Analytic Approaches for Policy Design

Collaborators: Saba Siddiki (Syracuse U), Seth Frey (UC Davis), Doug Rice (UMass Amherst), Edella Schlager (U of Arizona), Brenda Bushouse (UMass Amherst), Tanya Heikkila (UC Denver) and Chris Weible (UC Denver). 

Project website: institutionalgrammar.org

Research Questions: Is it possible to extract rules from rich digital text into a structured database based on a syntax for rules called the “Institutional Grammar”? If so, what quantitative metrics would institutional analysts want to be able to calculate based on the IG-database?

Norms and Rules – institutions — govern nearly everything humans do. Yet it is quite difficult to study the composition of rules, such as articulated in policy statements like regulations or simply operating in the minds of humans as norms, in any quantitative way. In this US NSF-funded project, we seek to design and develop computational methods that help the institutional analyst: (1) extract rules from rich text and into a structured database built on a syntax called the Institutional Grammar (IG); and (2) analyze, quantitatively and comparatively, rule configurations, using this database. Under this grant, I co-lead a multi-university “Research Coordination Network” (RCN) organized following commons-based peer production principles. Recently we’ve had some very exciting results that suggest that a machine learning approach is performing very well on the #1 task of extracting rules into IG-based syntax from rich text. Over the first year and a half, our RCN team has produced a variety of publications and has grown an international research group of nearly 70 scholars and graduate students worldwide. (More information available at https://institutionalgrammar.org/.) This effort is the beginning of what I expect to be a 10-20 year effort, but if the long term effort is successful, it will have huge implications for moving toward automated, comparative quantitative analysis of institutional designs in areas such as public policy (e.g., regulations), or anywhere else where norms and rules guide human behavior.

World Librarians

Collaborators: Carl Meyer and team (ShiftIT, Malawi), Waruks Wachuria and team (Net Bila Net, Kenya), Jeremy Smith (UMass), Scott McCullough (UMass ’19), Micky Cox (UMass ’21), Kate Marchetti (UMass ’21), Rachel Mann (UMass 21′), Grace Buckner (UMass ’21), and others. 

Website: https://worldlibrarians.org

One of my graduate school mentors said “Theory without experience is fantasy. Experience without theory is blind.” In my efforts to theorize about and empirically investigate instances of CBPP and Knowledge Commons (KC), I sought to implement some form of CBPP/KC effort on the ground to gain that real-world experience.  World Librarians (WL) is a program that I invented with my students and a partner organization in Malawi called ShiftIT, that allows me to gain this on-the-ground experience.

Specifically, World Librarians is a socio-technical system invented through a collaboration between ShiftIT in Malawi and students at UMass Amherst. World Librarians provides educational content to schools and libraries in Malawi and Kenya. These schools have no real library and have no access to the Internet, causing the children and library patrons to have very limited information to support their learning.

In Malawi we are operationally supporting over 20 schools. We also have a startup in Kenya with our partner Net Bila Net, reaching children living in a slum area outside of Nairobi. 

How does it work?

The schools and libraries we support have established some access to computer labs and a workable offline Wi-Fi router system called the RACHEL (learn more about the RACHEL device here). The UMass World Librarians team is charged with searching for digital educational content on topics that teachers and students request. Once uploaded, we rely on the teachers at these schools in the Global South to download the content for use in classrooms and computer labs, using their personal cell data plans. Using an app called World Remit, we can pay for the data used by the courier in downloading what we send through a simple and small credit-card payment. We call this “digital postage.” 

We are working to expand the program into Ghana, Haiti, and possibly Cameroon.

For more information:

Selected Publications