SIS Projects

Surprising Drawing Apprentice

by Nicholas Davis

Computer colleagues are a new type of artificial intelligent agent that acts as a partner in the creative process, i.e. contributing to a shared creative artifact with the user. This project seeks to design, develop, and evaluate a co-creative drawing partner that collaborates with the user in real time to draw surprising results on a shared canvas. The system will analyze the user’s drawing and respond in ways it predicts will surprise the user, such as 1) adding surprising features to objects the user has drawn, 2) adding surprising objects to the drawing based on recent contributions of the user, and 3) analyzing the entire canvas and determining what would be most surprising given the entire history of the drawing. The student should have a strong web-based programming background, both front-end and back-end development.

Understanding the Neurocognitive Dynamics of Co-Creation

by Nicholas Davis

Collaboration can inspire, motivate, and creatively engage individuals in a way that leads to more creative and diverse results than working independently. However, the neurological and cognitive mechanisms of collaboration are not well understood. This project seeks to quantify the collaboration of individuals engaged in collaborative drawing, as well as the neural mechanisms engaged during collaboration. We will perform observational studies investigating the collaboration strategies and interactions of individuals engaged in collaborative drawing in two domains: 1) collaboration between humans, and 2) collaboration between a human and a co-creative agent. These findings will be used to develop computational models of collaboration and co-creation for use in designing and evaluating co-creative agents.

Investigating the Neural and Collaboration Dynamics of Design Creativity

by Nicholas Davis

Design research investigates how individuals engage in the process of designing an artifact, such as identifying different stages of the design process and behaviors utilized to explore, refine, and ultimately implement design ideas. In this project, we seek to investigate how individuals work together to achieve a design solution, i.e. their collaboration dynamics. As part of this investigation we will analyze the neural activation patterns of individuals as they engage in a collaborative design task. This project will entail an observational study of collaborative design and analysis of data generated through that investigation.

The Institutional Context of CS Education

by Tonya Frevert

CS education is undergoing a paradigm shift as faculty consider how to best prepare CS majors for the demands of the 21st century workforce. Many traditional classroom lectures are giving way to the adoption of engaged teaching practices, such as flipped classrooms, lightweight teams, and active learning. While research has examined the effects of these practices on student performance outcomes, most studies are “one-off” reports within the context of a particular class in which X instructor at X university tries out new teaching practices and shares their singular experience with the CS education community. The institutional contexts around these efforts are seldom reported, let alone examined. This project will consider the effects of both the internal organizational environment on the use and effectiveness of CS teaching practices as well as issues stemming from the external social/political/economic structure that inevitably find their way into the classroom (e.g., inequality, public policy, employment). Participation in this research project involves collection and analysis of survey and interview data.

Extracting and representing knowledge in publications of clinical studies

by Yaorong Ge

Physicians currently rely on timely review of publications of clinical studies to maintain and update their knowledge for high quality patient care. However, as a part of the information explosion in recent decades, physicians are finding it difficult, if not impossible, to keep up with the latest clinical knowledge. One strategy to tackle this challenge is to transform knowledge in the narrative publications into computer models so that large scale knowledge integration and automatic reasoning become feasible. With the support of such models, physicians will be able to offer highest quality personalized patient care by applying the latest clinical knowledge that best fits the conditions of individual patients. This project aims to develop methods for knowledge extraction that focuses on one specialty of patient care, radiation treatment. Students will work with graduate students and postdoctoral researchers in SIS as well as in the Department of Radiation Oncology at Duke University Medical Center to analyze the content of radiation treatment publications, develop methods for knowledge extraction, and assess the performance of various models and modeling approaches. The student should have a strong background in a high level programming language (such as Java and/or Python). Knowledge of artificial intelligence, natural language processing, and/or health informatics is a plus.

Turtle Embroidery

by Celine Latulipe

This project involves investigating LOGO-turtle like drawing (which is a way to get young people involved in programming) as an interface to digital embroidery. This project involves creating a turlte-based programming/drawing environment that outputs a digital embroidery pattern. The student will have to deeply investigate turtle-like programming/drawing environments and the stitch patterns that digital embroidery machines use. Skills required include prototyping, design and strong web-based programming skills, both back-end and front-end.

An AI-based app for dietary behavior change

by Mary Lou Maher, Kaz Grace

As part of an NSF-funded project we are developing algorithms that can predict what will make a user curious. This project will design and develop a mobile app that applies these personalized curiosity algorithms to the domain of food and nutrition. More diverse diets have been shown to be healthier, and this app will encourage dietary diversity by suggesting new foods that our system predicts will inspire curiosity. This interdisciplinary project will combine elements of machine learning, human-centred design, recommender systems, and mobile development.

Pique: Encouraging Curiosity with AI

by Mary Lou Maher, Nadia Najjar, Xi (Sunshine) Niu

This project addresses a big data problem: how can search engines present information that people will find surprising and valuable? Building on research in recommender systems and serendipity, this project builds AI models to find and present information that will encourage curiosity. This project has various aspects the student can be involved. Tasks include the collection and analysis of data, model prototyping (curiosity/surprise), performance testing, and UROCFrevertevaluations with user studies that collect data from verbal protocols, interviews, eye tracking, EEG, and fNIR.

Summarizing Data Flows for Security

by Thomas Moyer

Securing systems against a well-equipped adversary is a daunting task, and often the adversary is able to get in despite the defenders’ best attempts to keep them out. One technology that can help swing the tide in favor of the defenders is data provenance. Provenance is the history of ownership and processing of data as it moves through and between systems, and is a vital component of building systems that are able to “fight-through” an attack. In building systems that are provenance-aware, often the gathered provenance is fine-grained and hard to reason about for an average user. To address this issue, techniques are developed that summarize the provenance in human-readable form. In this project, we will explore several summarization techniques and apply them to an existing provenance-aware system, extending the capabilities of those systems and making the provenance data easier to reason about.

BlockNet: Addressable Building Blocks

by Dimitrios Papanikolaou

What if buildings consisted of smart building components able to sense their structural, environmental, or human-imposed stimuli, communicate with each other, and reconfigure themselves like local internets? You will develop a proof-of-concept construction kit of interactive addressable blocks that once connected they can learn their assembly topology. The blocks will exchange data using a packet switching protocol, in which each block can turn into a transmitter, receiver, or router for messages, through physical contact. You will also explore application in building industry. Required skills: prototyping, digital electronics with Arduino, network communications, digital fabrication.

Viral intelligence in MoD systems

by Dimitrios Papanikolaou

What if intelligent mobility on demand (MoD) systems exchanged information virally, by using vehicles to physically carry data between stations, instead of using telecommunications? In this project you will explore the potential of MoD systems to distribute information through vehicles, similarly to how colonies of microorganisms communicate. Using agent based models of MoD systems in NetLogo, you will explore how information latency affects performance, and what relationship fleet size to trip pattern must have for a minimum desired latency. For example, stations with less frequent interactions with the rest of the network have more outdated perceptions of the system’s status and must therefore forecast over longer periods. Assuming that trip patterns that cyclic, we want to explore whether it is possible for a system to learn and develop a good estimate of its current state and whether fleet size plays a role in that. Required skills: multi-agent or agent-based modeling, machine learning, data mining.

SD.JS: System Dynamics JavaScript Library

by Dimitrios Papanikolaou

System dynamics is a methodology to macroscopically model dynamic systems using concepts from hydraulics and basic modeling blocks such as stocks and flows. Commercially available system dynamics software is expensive and difficult to customize. In this project we develop a JavaScript library for urban applications of system dynamics that allows researchers to model, simulate, and visualize systems directly in a web browser. You will curate the development of a modular library in JavaScript for system dynamics models and develop a series of demos/examples to illustrate each concept in a website. Required skills: JavaScript, D3.JS, HTML, CSS, and a basic knowledge of system dynamics.

Electronic MarketPlaces for Dynamically Priced MoD Systems

by Dimitrios Papanikolaou

Incentivizing users to rebalance vehicles instead of workers is an effective way to operate Mobility on Demand systems. We explore the design of a two-sided web-based marketplace for dynamically pricing trips in point-to-point MoD systems, such as car or bike sharing systems. Users “buy” vehicles from origin stations and “sell” them back to destination stations; trip values derive as transactional differences between buying and selling and can be positive, negative, or zero. You will help developing a marketplace dynamic web app, both frontend and backend, and a visualization interface, that allows two sides of online users (riders and traders) to transact over real stations on a map, in real time. The deliverable is a fully functional web-app that can be later used for conducting experiments. Requires expertise in data visualization and in dynamic web technologies such as Node.JS, Socket.IO, MongoDB, Leaflet.JS, JavaScript, and D3.JS.

Bidirectional Interactive Wooden Surfaces

by Dimitrios Papanikolaou

What if our physical environment could remotely mediate the traces of our interactions with it? We explore the capacity of interactive wooden surfaces to capture, manifest, and mediate traces of our bodily contacts with them, by sensing thermal imprints or vibrations and by replicating them in other wooden surfaces through the internet. You will develop a touch-sensitive thermochromic display matrix that can both sense and display bodyprints. The interface will consist of a layer of thermochromic paint and an underlying layer with a matrix of Peltier thermoelectric modules. The Peltier modules can act both as temperature sensors and as output heating actuators which in turn will change the pigment of the thermochromic paint. By connecting an array of Peltier modules to a microcontroller and to the internet you will explore application scenarios, prototyping methods, and performance tests. Requires prototyping, engineering, electronics, programming, and design skills.

Hybrid Mobile App Security

by Meera Sridhar

This project will explore programming language-based security techniques to address critical security and privacy issues in hybrid mobile apps. Hybrid app frameworks allow mobile developers to design app code using web technologies alone, and supply native and bridge (APIs for accessing device resources) code necessary for instant porting to several mobile platforms. The last few years have seen an explosive growth in the share of hybrid mobile apps worldwide, coinciding with the increasing ubiquity of HTML5. The main goal of the research is to develop technologies to mitigate dangerous security and privacy vulnerabilities in hybrid app frameworks, by creating an airtight security model robust against bridge exploits, and minimizing the overall attack surface of the hybrid framework. We will use binary hardening techniques to provide an elegant and effective solution towards this end goal. The research will provide automatic protection of PhoneGap Android hybrid apps, transparent enforcement that preserves the functionality of safe apps, and retroactive enforcement that can be applied to vulnerable PhoneGap Android apps in the wild. The student will help create a testbed of vulnerable hybrid apps and attacks. He/she will also help with building the binary hardening engine. The required skill set includes strong programming and scripting skills, some cyber security background or a desire to learn about security; knowledge of Android and/or JavaScript is a plus.

Study of Advanced Persistent Threats

by Jinpeng Wei

This project will explore novel techniques that can be used by APT (Advanced Persistent Threat) attacks and develop effective countermeasures. It involves collecting real-world malware samples and threat intelligence about them from the Internet, and analyzing the artifacts using techniques such as threat modeling and machine learning. The student will crawl the web for useful information and then process the information to infer useful patterns. Required skills include strong web programming, scripting, Java programming, statistical modeling, and logic programming (such as Prolog). Please apply by September 15, 2017.

Making Sense

by David Wilson, Johanna Okerlund

As Human-Computer Interaction researchers, we are interested in studying the human experiences that surround novel and emerging technologies. Who benefits from the technology? What are their motivations for using it? What are the social interactions that happen around and as a result of the technology? Does the technology reinforce existing inequities or does it have the potential to disrupt them? We ask these types of questions in relation to the CCI Makerspace, a new open-ended space for UNCC community members to work on self-directed and collaborative projects with the help of digital fabrication and traditional making tools. We have reason to believe that rich learning experiences happen as a result of participating in the space, but we need better ways to capture and understand these experiences. This research project investigates the pathways of people who participate in the space over short and extended periods of time: What are their evolving short term and long term goals? How do they spend their time in the space? Who do they interact with? How do their perspectives, attitudes, and identity change? Participation in this research project involves the collection and analysis of data in the form of surveys, interviews, and traffic logs to help answer these questions.