Theses, Dissertations, and other Required Graduate Degree Essays

Receive updates for this collection

This collection contains digitized SFU theses except for those theses submitted within the last 12 months. If you cannot find the thesis you are looking for please search Recently Submitted Theses as it may be a recently submitted thesis and thus not yet available in Summit.

ChatrEx: Designing explainable chatbot interfaces for enhancing usefulness, transparency, and trust

Author: 
Date created: 
2021-07-16
Abstract: 

When breakdowns occur during a human-chatbot conversation, the lack of transparency and the “black-box” nature of task-oriented chatbots can make it difficult for end users to understand what went wrong and why. Inspired by recent HCI research on explainable AI solutions, we explored the design space of explainable chatbot interfaces through ChatrEx. We followed the iterative design and prototyping approach and designed two novel in-application chatbot interfaces (ChatrEx-VINC and ChatrEx-VST) that provide visual example-based step-by-step explanations about the underlying working of a chatbot during a breakdown. ChatrEx-VINC provides visual example-based step-by-step explanations in-context of the chat window whereas ChatrEx-VST provides explanations as a visual tour overlaid on the application interface. Our formative study with 11 participants elicited informal user feedback to help us iterate on our design ideas at each of the design and ideation phases and we implemented our final designs as web-based interactive chatbots for complex spreadsheet tasks. We conducted an observational study with 14 participants to compare our designs with current state-of-the-art chatbot interfaces and assessed their strengths and weaknesses. We found that visual explanations in both ChatrEx-VINC and ChatrEx-VST enhanced users’ understanding of the reasons for a conversational breakdown and improved users' perceptions of usefulness, transparency, and trust. We identify several opportunities for future HCI research to exploit explainable chatbot interfaces and better support human-chatbot interaction.

Document type: 
Thesis
File(s): 
Supervisor(s): 
Parmit Chilana
Department: 
Applied Sciences: School of Computing Science
Thesis type: 
(Thesis) M.Sc.

Converter based electrochemical impedance spectroscopy for fuel cell stacks

Author: 
Date created: 
2021-03-19
Abstract: 

Fuel cells are important devices in a hydrogen-based chain of energy conversion. They have distinctive advantages over batteries with their higher energy density and faster refueling speed, which make them attractive in stationary power supplies and heavy-duty vehicles. However, the high cost and low durability associated with modern fuel cells are still hindering their wider commercialization. Besides developing more reliable and lower cost materials and advanced assemblies of cells and stacks, a practical and effective diagnostic tool is highly needed for fuel cells to identify any abnormal internal conditions and assist with maintenance scheduling or application of on-board mitigating schemes. Conventionally, linear instruments were used for fuel cell EIS, however, limited to single cells or short stacks only as a laboratory testing method. With recent developments, EIS enabled by switching power converters are capable of being applied to a high-power stack directly. This approach has the potential for practical field applications such as a servicing tool for fuel cell manufacturers or an on-board diagnostic tool of a moving vehicle. Previous works on converter based EIS have made a few different attempts at conceptually realizing this solution while several significant issues were not well recognized and resolved yet. As such, this thesis explores further on this topic to address the flexibility of EIS perturbation generation, the perturbation frequency range, and the linkage between fuel cell EIS requirements and the converter design to push for its readiness for practical implementations. Several new solutions are proposed and discussed in detail, including a total software approach for existing high-power converters to enable wide-frequency-range EIS, a redesign of the main dc/dc converter enabling wide-frequency-range perturbations, and a separate auxiliary converter as a standalone module for EIS operation. A detailed analysis of oscillations brought by converter based EIS in powertrains is also presented.

Document type: 
Thesis
File(s): 
Supervisor(s): 
Jiacheng Wang
Department: 
Applied Sciences: School of Mechatronic Systems Engineering
Thesis type: 
(Thesis) Ph.D.

Sensemaking with learning analytics visualizations: Investigating dashboard comprehension and effects on learning strategy

Author: 
Date created: 
2021-04-07
Abstract: 

In the provision of just-in-time feedback, student-facing learning analytics dashboards (LADs) are meant to aid decision-making during the process of learning. Unlike summative feedback received at its conclusion, this formative feedback may help learners pivot their learning strategies while still engaged in the learning activity. To turn this feedback into actionable insights however, learners must understand LADs well enough to make accurate judgements of learning with them. For these learners, LADs could become an integral part of their self-regulatory learning strategy. This dissertation presents a multifaceted examination of learners’ sensemaking processes with LADs designed to support self-regulatory learning. The in-situ studies detailed therein examine learners’ understanding of the data visualized in LADs and the effects of this understanding on their performance-related mental models. Trace data, surveys, semi-structured in-depth qualitative interviews, and retrospective cued recall methods were used to identify why, when, and how learners used LADs to guide their learning. Learners’ qualitative accounts of their experience explained and contextualized the quantitative data collected from the observed activities. Learners preferred less complex LADs, finding them more useful and aesthetically appealing, despite lower gist recall with simpler visualizations. During an early investigation of how LADs were used to make learning judgments in situ, we observed learners’ tendency to act upon brief LAD interactions. This inspired us to operationalize gist as a form of measurement, describing learners’ ability to make sense of a LAD after a brief visual interrogation. Subsequent comparisons of the accuracy and descriptiveness of learners’ gist estimates to those of laypeople repeatedly showed that laypeople were more apt than learners to produce accurate and complete gist descriptions. This dissertation culminates in a final study examining the evolution of learners’ mental models of their performance due to repeated LAD interaction, followed by a discussion of the contextual factors that contributed to what was observed. Trends observed across this work suggest that learners were more apt to “get the gist” with LAD after repeated interaction. This dissertation contributes a novel method for evaluating learners’ interpretation of LADs, while our findings offer insight into how LADs shape learners’ sensemaking processes.

Document type: 
Thesis
File(s): 
Supervisor(s): 
Marek Hatala
Department: 
Communication, Art & Technology: School of Interactive Arts and Technology
Thesis type: 
(Thesis) Ph.D.

The effects of sediment organic carbon and chemical residence time on the acute toxicity of sea lice chemotherapeutants to benthic invertebrates

Author: 
Date created: 
2021-03-25
Abstract: 

Chemotherapeutants are commonly used to manage sea lice outbreaks in salmonid aquaculture. Among the classes of chemotherapeutants used are avermectins; these tend to persist in the sediments underneath salmon farms and may directly impact nearby benthic fauna of marine ecosystems. The present study sought to determine how two environmental factors – namely, sediment organic carbon (OC) and chemical residence time – can modify the toxicity of emamectin benzoate (EB; formulation: Slice®) and ivermectin (IVM) in two species of benthic invertebrates: the amphipod Eohaustorius estuarius and the polychaete Neanthes virens. In both species, sediment OC significantly reduced toxicity, an effect that was more pronounced for IVM and combination exposures. Four months of chemical residence time reduced toxicity in E. estuarius but did not affect toxicity in N. virens. This research provided novel insight into the effects of two environmental factors that potentially impact avermectin toxicity in nontarget species underneath salmon farms.

Document type: 
Graduating extended essay / Research project
Supervisor(s): 
Chris Kennedy
Department: 
Science: Biological Sciences Department
Thesis type: 
(Project) M.E.T.

Evaluating long short-term memory networks for modeling land cover change

Date created: 
2019-08-14
Abstract: 

Land cover change (LCC) can be viewed as dynamic complex systems which require relevant relationships to be encoded when represented within various modeling approaches. Recurrent Neural Networks (RNNs), specifically the Long Short-Term Memory (LSTM) variant, belong to a category of Deep Learning (DL) approaches best suited for sequential and timeseries data analysis, thus suitable for representing LCC. The primary objective of this study is to examine the capacity and effectiveness of LSTM networks for forecasting LCC given varying geospatial input datasets with feature impurities. Using synthetic and MODIS land cover datasets for British Columbia, Canada, results demonstrate the sensitivity of LSTM models to varying geospatial input dataset characteristics. Geospatial datasets with finer temporal resolutions and increased timesteps yielded favourable results while coarser temporal resolutions and fewer timesteps were affiliated with less successful outcomes. This thesis research contributes to the advancement of automated, data-driven DL methodologies for forecasting LCC.

Document type: 
Thesis
File(s): 
Supervisor(s): 
Suzana Dragicevic
Department: 
Environment: Department of Geography
Thesis type: 
(Thesis) M.Sc.

Quantifying carbon cycle feedbacks under negative CO2 emissions

Date created: 
2021-07-09
Abstract: 

Land and ocean carbon sinks play a major role in regulating atmospheric CO2 concentration and climate. However, their future efficiency depends on feedbacks in response to changes in atmospheric CO2 concentration and climate. Since negative CO2 emissions technologies (NETs) are a key mitigation measure in emission scenarios consistent with global climate targets, understanding carbon cycle feedbacks under negative CO2 emissions is essential. This thesis investigates carbon cycle feedbacks under positive and negative CO2 emissions using an Earth system model driven with idealized scenarios of increasing and decreasing atmospheric CO2 concentration. Results suggest that carbon cycle feedbacks differ under positive and negative emissions, independently of the specific approach chosen for their quantification. The findings of this thesis provide insights into the approach best suited to quantify carbon cycle feedbacks under negative CO2 emissions, and into the role of these feedbacks in determining the effectiveness of NETs in reducing CO2 levels.

Document type: 
Thesis
File(s): 
Supervisor(s): 
Kirsten Zickfeld
Department: 
Environment: Department of Geography
Thesis type: 
(Thesis) M.Sc.

Mushroom for improvement: a model for the circulation of fanfiction sub-genres

Date created: 
2021-06-28
Abstract: 

This thesis explores the circulation of fanfiction sub-genres across fan communities as starting point for further inquiries into fan object movement among fans. Fan studies has long been interested in the circulation of fan objects, but lacks a broad understanding of how these objects move through space and time (Hills 2014). In applying Mikhail Bakhtin’s concept of heteroglossia to describe fan communities, objects and circulatory behaviours, I analyze two case studies to propose a new model. The first tracks the circulation of a trope on Tumblr, while the second explores the movement of a fanfiction sub-genre across platforms, post types and fandoms. My proposed model is based on the radiating structure of mycelium (the vegetative part of a fungus). Mycelium’s branching and agile nature provides a more accurate framework for ever-evolving fannish circulatory practice.

Document type: 
Graduating extended essay / Research project
File(s): 
Supervisor(s): 
Hannah McGregor
Department: 
Communication, Art & Technology: Publishing Program
Thesis type: 
(Project) M.Pub.

The applications of interaction between light and chemical reactivity in molecular systems

Author: 
Date created: 
2021-06-30
Abstract: 

This thesis presents a series of scientific examples that provide further insight into the connection of light and chemical reactivity for applications in catalysis, targeted drug delivery, and detection. Small organic molecules based on 1,2-dithienylethenes (DTEs) and noble metal-based nano-assemblies were incorporated into the system to impart control over their physical and chemical properties, and activity. The light was used as an external stimulus to initiate chemical reactions by causing either a structural, electronic, or chemical change in the chromophore or energy transformation (such as heat) in the system. In the first example (Chapter 2), light is employed to generate the variation in the activity of Karstedt’s catalyst by introducing a DTE-based inhibitor. It is demonstrated that the electronic changes that occur during the light-assisted isomerization of dithienylethene lead to changes in the binding strength of the inhibitor to the catalyst’s metal centre. This, in turn, leads to changes in the reaction progression of simple hydrosilylation reactions. In the second example (Chapter 3), a novel dual-mode (colourimetric and fluorescent) optical probe based on photoresponsive dithienylethene is developed for fast and user-friendly detection and consumption of hydrazine in the vapour and solution phase. The molecule reacts with hydrazine to induce an irreversible structural change in the 1,3,5-hexatriene system leading to deactivation of photochromic activity by quenching effects and simultaneously generating an emissive response. In the third example (Chapter 4), light is used as an external stimulus to indirectly gate the chemical reactivity for an on-demand release of therapeutics in aqueous environments by creating nanoassemblies based on gold nanoparticle core. The constructed nanoassemblies consist of the thermoresponsive Diels-Alder adducts of the molecules to be released close to gold nanoparticles enclosed in an amphiphilic polymer shell. The system uses the photothermal effect of gold nanoparticles to trigger thermo-responsive retro Diels-Alder reactions in the vicinity of nanoparticles to release therapeutic molecules in aqueous solutions.

Document type: 
Thesis
File(s): 
Supervisor(s): 
Neil Branda
Department: 
Science: Department of Chemistry
Thesis type: 
(Thesis) Ph.D.

Protecting privacy of semantic trajectory

Author: 
Date created: 
2021-06-29
Abstract: 

The growing ubiquity of GPS-enabled devices in everyday life has made large-scale collection of trajectories feasible, providing ever-growing opportunities for human movement analysis. However, publishing this vulnerable data is accompanied by increasing concerns about individuals’ geoprivacy. This thesis has two objectives: (1) propose a privacy protection framework for semantic trajectories and (2) develop a Python toolbox in ArcGIS Pro environment for non-expert users to enable them to anonymize trajectory data. The former aims to prevent users’ re-identification when knowing the important locations or any random spatiotemporal points of users by swapping their important locations to new locations with the same semantics and unlinking the users from their trajectories. This is accomplished by converting GPS points into sequences of visited meaningful locations and moves and integrating several anonymization techniques. The second component of this thesis implements privacy protection in a way that even users without deep knowledge of anonymization and coding skills can anonymize their data by offering an all-in-one toolbox. By proposing and implementing this framework and toolbox, we hope that trajectory privacy is better protected in research.

Document type: 
Thesis
File(s): 
Supervisor(s): 
Nadine Schuurman
Department: 
Environment: Department of Geography
Thesis type: 
(Thesis) M.Sc.

The web publication development process for engaging small communities

Author: 
Date created: 
2021-04-27
Abstract: 

This report outlines the process used for creating the Master of Publishing 25th anniversary web publication and discusses the design and editorial practices implemented to engage the main audience, the program’s alumni. The report also explores the application of Lean manufacturing, a process improvement methodology, to the publication development process and suggests Lean practices and tools that can be used to improve and maintain the publication in the future. Best practices for the management of the publication and increasing community engagement are also discussed.

Document type: 
Graduating extended essay / Research project
File(s): 
Supervisor(s): 
Mauve Pagé
Department: 
Communication, Art & Technology: Publishing Program
Thesis type: 
(Project) M.Pub.