The way cities handle and process public feedback is manual, resource-intensive, and fails to leverage advancing technology, reinforcing the United Nations’ call for sustainable cities and strong institutions (SDGs 11, 16). To address these needs, Illume seeks to empower civic engagement professionals to make informed decisions with confidence. Through this web application, complex qualitative data is transformed into clarity, employing a human-in-the-loop process to ensure step-by-step traceability and trustworthy analysis for the user.

A big thanks to our advisor, Morgan Boyco RPP MCIP, for his guidance and insights on our project. Also, a shoutout to the interdisciplinary capstone teaching team, the many urban planning students with whom we interacted, and any other stakeholders who took the time to converse with us, whether synchronously or asynchronously.

During my engineering undergraduate at the University of Waterloo, I was in a two-term interdisciplinary capstone course with my two friends. As a team, we were drawn to how cities pursuing sustainability should leverage technological advancements in their planning processes. From this, we conducted research and conversations with stakeholders in academia and industry, allowing us to define three significant challenges faced in the analysis of civic engagement data.

First, the current analysis process is extremely time and resource-intensive. Second, there is an ever-growing need for trust, explainability, and ethical responsibility in the public sector. And third, cities and planning professionals struggle to take advantage of new and innovative technical tools.

Illume is a web application that transforms the complexity of civic engagement data into clarity, empowering users to make informed decisions with confidence. Existing digital tools are often overwhelming—they are complex and opaque, leaving users unable to explain and trust where the results are coming from. In civic engagement, where public trust in the decision-making process is paramount, this lack of transparency worsens the problem.

Illume changes that. The first key feature is thematic analysis, a well-established evaluation method, that buckets data into different themes. This helps the user to quickly evaluate their dataset. The second key feature is integrating a human-in-the-loop process, incorporating the unique human expertise and contextual awareness of the user to then accept or reject the given classification. In the end, this creates a high quality, ground truth dataset that can be used for further analysis. With this process, users can feel confident and be able to articulate how they used Illume in their work.

To outline where Illume is situated in the larger planning process, let's first overview of what the planning process looks like and why there was a reason for Illume in the first place. I'll first introduce what qualitative data analysis entails. Qualitative Data Analysis (QDA) is very common in the social sciences, but less so in the engineering discipline. There are many ways to explain QDA, and we do not assert that this particular way is the only way to understand it. We refer to (Li, 2022) for a general approach to QDA. At a high level, these are the four steps:

1. Type of Data → What you’re working with: observations, surveys, interviews, transcripts, focus groups, documents, audio, etc.
2. Analysis Approach → Guided by theory: Inductive (grounded theory, build new ideas from scratch). Deductive (test/apply existing theories). Tools include keyword searches, conversation analysis, semiotics, word counts, etc.
3. Processing the Data → Apply methods like: Coding (open → axial → selective). Memoing (notes, concept maps, storylines).
4. Communicating Results → Tailor insights to the audience.

With this understanding of QDA in mind, we turn to the planning process. The full process has 9 steps (from identifying issues → implementing programs). But for Illume’s context, we can boil it down to 3 essentials:

1. Problem Identification & Objectives → Define what matters.
2. Data Collection & Interpretation → Gather and make sense of information.
3. Solution Evaluation & Implementation → Test ideas, refine, and put them into practice.

In the planning process, qualitative data analysis (QDA) takes place during the stage of data collection and interpretation, where planners identify the types of data they have, choose an analysis approach, process that data through methods like coding, and finally communicate the results. Illume fits into this workflow by focusing specifically on textual survey responses in .csv format. Its approach is inductive, surfacing broad, aggregate-level themes—the “forest” rather than the “trees.” Through open coding, it helps identify patterns across responses, but it does not extend into the communication stage, leaving that task to planners. In short, Illume targets step two of the planning process and, within QDA, concentrates on steps two and three: analysis and coding.

To ground this problem in lived experience, we developed three personas that illustrate how different stakeholders interact with civic engagement data and where current processes fall short. These personas served as guiding anchors in our design process, ensuring our objectives stayed human-centered and relevant:

Alongside the double diamond design process described earlier, we also needed a clear methodology for gathering data. In the early ideation stages (Spring 2024), we cast a wide net. We conducted exploratory interviews with professors in sociology, legal studies, English, and communication studies, who highlighted the importance of equitable technology. We reached out to community groups (e.g., literacy and shelter organizations) to understand real-world needs, and engaged with urban planning students who introduced the idea of Wicked Problems. Classroom discussions, including those with the Future Cities Institute, further shaped our thinking and ultimately led us to focus on the urban planning space.

Building on the research and methodology outlined above, we developed Illume as our response to the challenges of civic engagement data analysis. Illume is a web application that turns complexity into clarity, empowering users to make informed decisions with confidence.

At its core, Illume is guided by two principles:

1. Efficiency through automation: accelerating the manual, resource-intensive process of qualitative analysis.
2. Trustworthiness through transparency: ensuring results are interpretable and that professional judgment is never replaced.

Our design philosophy is human-centered: Illume augments the expertise of planners rather than substituting for it. By streamlining repetitive coding work, the tool frees users to focus on interpreting insights and applying them strategically.

1. Thematic Analysis

After the data is uploaded, the key feature of Illume—thematic analysis—is applied. Thematic analysis is a proven methodology in qualitative research that transforms unstructured public feedback into meaningful, organized categories. This type of analysis was elaborated on in Section 2.1, under the problem context.

By applying natural language processing to civic engagement data, Illume can efficiently process hundreds of public comments in under 20 seconds. For example, Illume automatically groups together concerns about “parking” even if respondents phrase it differently, such as “not enough spaces” or “nowhere to park.” In our sample dataset, we used actual public feedback from the Stage 2 ION proposal at the Region of Waterloo (see Appendix D).

A benefit Illume offers is consistency, applying the same framework to every response, compared to humans who may vary in approach (even due to mood). While this consistency is valuable, we recognize that AI systems are not immune to bias or error. This is why the human-in-the-loop process (discussed in Section 6.2.2) is critical for oversight and transparency, building on the suggestions of Borchers et al. Ultimately, this automated step reduces the manual effort of sorting comments into themes, allowing professionals to focus on higher-value analysis.

In Illume, this oversight occurs twice: first, when the user accepts or rejects whether a comment belongs in a suggested theme; and second, on the rejection screen, when the user decides if the rejected comment should be reassigned to another theme. Screenshots of this process are provided in Appendix B. In practice, this means the planner still reads through each comment, applying their expertise just as they would in traditional thematic analysis.

As development began, we shifted to more targeted stakeholder interviews. In our co-op and Winter 2025 terms, we tested Illume prototypes directly with professional planners and urban planning students. These conversations—some formal, some more exploratory “research conversations”—gave us iterative feedback for refining the product. Internally, we supported this process with established engineering practices: two-week sprints, Jira for task tracking, version control, and regular sync meetings. These practices ensured consistent progress while leaving room for iterative input from stakeholders. Finally, as part of solution verification, we conducted eight semi-structured user interviews. These allowed us to test our assumptions, evaluate Illume’s potential fit, and gather qualitative feedback. Since this project was not intended for formal publication, our work did not require ethics clearance from the Office of Research Ethics, though all team members had completed TCPS-2 CORE and followed ethical principles throughout.

To evaluate Illume more systematically, we grounded our interviews in the Technology Acceptance Model (TAM) (Davis, 1986). TAM is widely used in Human–Computer Interaction (HCI) to assess user perceptions of new technologies, focusing on three core constructs:

1. Perceived usefulness
2. Perceived ease of use
3. Intention to use

Interview Design

We adopted a semi-structured format (Lazar et al., 2017), balancing consistency with flexibility. The structure was:

1. Introduce purpose and obtain consent.
2. Collect demographic/control questions.
3. First set of baseline statements (about current workflows and digital tool comfort).
4. Live walkthrough of Illume.
5. Second set of statements (TAM-based, using a 7-point Likert scale).
6. Open-ended qualitative questions + participant questions.

Before demonstrating Illume, we asked participants about their current workflows and comfort with technology. As shown in the figure below, most participants found their current analysis process of civic engagement data tedious, with 6 out of 8 participants either somewhat or moderately agreeing with this statement.

Baseline Statements

Confidence in current analysis accuracy was mixed. While most somewhat agreed they were confident, several raised concerns about potential bias or declining concentration when processing large volumes of responses. Participants were generally comfortable using digital tools, but reported notably less comfort with AI, highlighting an adoption barrier that Illume addresses through its human-in-the-loop approach.

1. I find the current analysis process of civic engagement data tedious.
2. I am confident in the accuracy of my analysis conclusions.
3. I am comfortable with using digital tools for analysis.
4. I am comfortable with using artificial intelligence (AI) for analysis.
5. I am confident that I could integrate some kind of AI tool into my analysis workflow.

TAM-Based Statements (Post-Demo)

After the Illume demonstration, participants showed a marked shift in their understanding of how AI fits within thematic analysis. As shown in the figure below , the majority strongly agreed that Illume was easy to use and would make their analysis process faster.

1. I found Illume easy to use. (perceived ease of use)
2. I would be able to explain where Illume was used in my workflow. (ease of use, usefulness)
3. I would be able to explain why Illume was used in my workflow. (ease of use, usefulness)
4. I am confident in the accuracy of Illume. (usefulness)
5. I am confident in being able to integrate Illume into my existing workflows. (ease of use)
6. I think Illume would make my current analysis process faster. (usefulness)
7. I would use Illume in my workflow. (intention to use)

We concluded our FYDP journey with an award-winning presentation, showcasing a working platform hosted on Vercel. This milestone reflected not only technical achievement, but also the impact and relevance of the problem Illume set out to address. It is fitting to end with the same framework that began this report: the double-diamond design process. In the problem space, we identified a significant challenge in urban planning—the manual, resource-intensive analysis of civic engagement data that fails to leverage advancing technologies. Through interviews and collaboration with planning experts, we gained insights into a field that was initially unfamiliar, allowing us to refine and validate our problem statement.

Transitioning into the solution space, we developed the first iteration of Illume. This tool focuses on thematic analysis of textual civic engagement data, paired with a human-in-the-loop verification process that empowers users with granular control over their analysis. Our validation interviews with planners and students confirmed Illume’s usability, usefulness, and transparency. Participants not only valued the time savings, but also expressed confidence that they could clearly explain where and why the tool was used in their workflows.

Beyond the double-diamond, Illume aligns with broader global frameworks. From the UN’s Sustainable Development Goal 11 (Sustainable Cities and Communities) to SDG 16 (Peace, Justice, and Strong Institutions), Illume contributes—albeit in a small way— to a shared future where civic technologies support trust, inclusivity, and effective governance.

Ultimately, Illume is more than a prototype. It is a glimpse of the future of civic intelligence— one where technology assists, but never replaces, human expertise in shaping the communities we live in.

To recap the alignment between identified problems and our solution, our research highlighted several key challenges in the civic engagement data analysis process. Each of these is directly addressed by Illume’s core features:

• Challenge: Manual, resource-intensive analysis.
  Illume’s Solution: Thematic analysis reduces initial processing time by 50–80% through automated categorization.
• Challenge: Concerns about trust and transparency in AI tools.
  Illume’s Solution: Human-in-the-loop validation ensures accuracy while maintaining essential oversight and judgment.
• Challenge: Difficulty explaining how conclusions were reached.
  Illume’s Solution: Step-by-step traceability documents each decision for complete transparency.
• Challenge: Existing digital tools overwhelm users with complexity.
  Illume’s Solution: User-centered interface simplifies the workflow with an intuitive design focused on clarity.
• Challenge: Fear that AI will replace human expertise.
  Illume’s Solution: AI is deliberately positioned as an assistant, preserving the critical role of human insight in the analysis process.

Through our interviews, we identified several priority areas for future development and consideration. These include, but are not limited to:

• Standardization: Establishing consistent data analysis procedures within planning teams and across municipalities.
• Customization: Adding more user-driven features such as exploratory visualizations and collaborative tools.
• Privacy & Ethics: Building privacy-by-design features, ensuring compliance with MFIPPA and the Canadian Institute of Planners (CIP) professional code of conduct. This includes removing personally identifiable information (PII) and addressing other potential sources of bias.