Bibliometrics+ is an AI- and EDI-driven library usage analytics system developed to support more transparent, data-informed, and equity-aware analysis in public library environments. At this stage of the project, the system integrates open datasets from Ottawa Public Library, Toronto Public Library, and Bibliothèques de Montréal into a unified analytical environment built on a Python-based ETL pipeline, a Supabase PostgreSQL backend, and a multi-page Streamlit dashboard. The completed prototype supports cross-city and branch-level KPI exploration, contextual EDI-related analysis, data-quality auditing, geographic library search, and an early-stage grounded AI explanation layer that interprets filtered dashboard outputs without generating unsupported claims. Overall, this report demonstrates that the project has progressed from conceptual planning and partial implementation into a functioning end-to-end prototype while also identifying the remaining limitations in cross-city comparability, formal testing depth, and AI maturity.

BuildNet-CA is a tool for municipal transparency created to overcome problems associated with a lack of integrated and easily accessible system for tracking housing and infrastructure development initiatives in Ottawa. It is difficult for residents, scholars, and all other stakeholders to understand what is going on with the city's current and future development projects as the information is currently scattered across multiple government departments and systems. From a simple idea, BuildNet-CA evolved into a comprehensive web-based solution in the course of this capstone project. The program utilizes the geographic visualization technology through an interactive mapping interface where the data collected from multiple government departments in a form of standardized database is presented. It becomes possible for users to analyze and make sense out of development information through the use of geographic visualization, a standardized set of data, and cryptographic authentication.

The IRM4900 Capstone course requires students and professors to coordinate multiple deadlines, reports, presentations, and evaluations across two terms. Currently, important information such as deliverables, deadlines, supervisor availability, and resource links is scattered across Brightspace. This increases the likelihood of miscommunication or missed requirements. Professors also face the repeated burden of clarifying schedules, responding to availability requests, and ensuring students stay on track. Beyond project management, there is also a gap in connecting capstone work with career development. Students often struggle to identify which skills their projects highlight, while professors have limited time and resources to stay updated on the constantly changing industry needs. With multiple courses and responsibilities, it can be difficult for supervisors to keep track of which skills are currently in demand. As a result, there is still no simple and convenient way for students or professors to see which capstone projects to develop and relate to real-world job skills and opportunities. The proposed Capstone Companion Dashboard addresses these challenges by providing a centralized platform for both students and professors. It will include deadline tracking, progress monitoring, supervisor availability, resource integration, and accessibility features. Additionally, an AI chatbot will analyze uploaded projects to identify demonstrated skills, recommend potential career paths, and highlight in-demand skills from job postings. This tool will help manage the capstone process while connecting academic work to real-world opportunities for both students and supervisors.

Chatalog is a chatbot that has been developed for libraries offering an intuitive, conversational search method for books using natural language. There is a lack of recommendation systems built into library websites; this tool will improve access to resources from its catalog for users. Through its services and deployment, Chatalog showcases an advanced use case for artificial intelligence, information management, and system design with its practical integration of knowledge graphs and large language models.

Due to the recent and ongoing advancements in artificial intelligence, large language models are becoming ubiquitous in society, shaping how people create and access information. Unfortunately, these deep neural networks can carry explicit and implicit biases learned from the gargantuan amounts of data they are trained on. When this biased information is used, its propagation can have negative effects on those affected and on society as a whole. As women and people of colour, tackling this problem means bringing awareness to information that may harmfully affect us. We will develop FairLLM to be a tool to help users identify biases and provide developers with a bias detection tool for their LLMs, thus promoting safer LLM usage and creation. Before creating the tool, we conducted a literature review to identify existing research on bias detection, mitigating bias and fairness while using LLMs. After reviewing 30 research articles, we conducted a comparative analysis to compare findings and identify what was missing to prevent the spread of misinformation for everyday LLM users. Based on our comparative analysis, we discovered that proposed metrics and frameworks can identify and lessen bias, but none propose a single, straightforward platform. The methods and frameworks from these studies do not allow everyday LLM users to evaluate responses. Nor do they aim to reduce the amount of bias returned from the LLM based on the user’s prompt. We found many frameworks employ the use of another LLM, creating the possibility of missed bias if biased themselves. Most methods also omit the evaluation of the user’s prompt that determines the LLM’s response. Research emphasis is also placed on certain social biases, leaving some unexplored. By combining fairness measures, bias avoidance and assessment methods, FairLLM seeks to close this gap as an innovative LLM bias discovery tool.

Library patrons often face challenges in wayfinding, especially in large academic settings where catalogues and floor layouts can be difficult to navigate. Traditional signage and digital catalogues alone are usually insufficient, which has led to growing interest in technology-driven solutions. AR has emerged as a promising tool for improving navigation. Studies show that AR can overlay helpful cues, reduce disorientation, and improve situational awareness, though most recommend pilot testing in specific contexts before widespread adoption. However, AR navigation is not without challenges. Tracking accuracy varies across devices, sometimes leading to orientation drift and reduced reliability. Low-cost alternatives such as QR code anchors have been proposed to help re-establish location indoors and maintain accuracy without expensive infrastructure. At the same time, reliable catalogue management remains critical. MARC 21 continues to be the dominant standard for bibliographic records, and research in library user experience emphasizes simplifying interfaces and conducting usability testing to ensure systems remain intuitive. Accessibility is another key requirement. The latest Web Content Accessibility Guidelines highlight mobile and touch interactions, such as ensuring target size and clear focus states, which are highly relevant for a patron-facing mobile navigation tool. Taken together, the literature supports integrating MARC-based cataloguing with AR-assisted or QR-anchored navigation, while prioritizing usability and accessibility testing. This approach balances innovation with reliability and inclusivity.

The main objective of our project was to conduct a research study, centred around how students perceive and use AI. With the rapid advances in AI, the education sector has experienced significant changes in students’ behaviour and academic performance. This has impacted students’ skills in the academic and professional environment. This phenomenon raises major concerns for academic institutions and among students, as AI tools are increasingly relied upon, potentially taking away the necessary experiences that students are meant to have while completing their studies. When used accordingly, AI may offer benefits to students, however, without guidance, some may find themselves using it unethically.

UniChat is a governance-aware Retrieval-Augmented Generation (RAG) platform designed for university environments, providing grounded, citation-backed answers from institution-approved content. This progress report summarizes team contributions, the refined problem statement, system architecture and workflow, current implementation status, experimental evaluation approach, governance considerations, and the roadmap to the final deliverables.

This Capstone project is based on the concept of "usability and professional standards" in relation to the current Carleton Counselling Services website. The current website does not appear to be adhering to the current Carleton standards in relation to the presentation of the website, nor does it seem to be following logical usability principles. This, in turn, has created a situation wherein the usability of the website has been compromised. In order to better improve the usability of our website, we will conduct user studies where they will perform certain tasks on our website and cross examine the results with the results gained from users doing the same tasks in the current Carleton Counseling Services website. This will enable us to compare the usability of the current website in relation to the prototype. The purpose of the project is to improve the usability of the Carleton Counselling Services website while adhering to the current Carleton standards in relation to the presentation of the website, which will enable the users to access the services offered in a better manner. In terms of visualization, it is directly linked to the usability of the website. We followed the guidelines provided by the standards set by Carleton University for websites while also relying on the main Carleton University website as further reference. The current Carleton Counselling Services website does not appear to adhere to these standards in relation to the presentation of the website, which are reflective of the professional nature of the university.