Recommender systems play a pivotal role in helping users navigate a vast selection of products and services. On online platforms, users have the opportunity to share feedback in various modes, such as numerical ratings, textual reviews, and likes/dislikes. Traditional recommendation systems rely on users’ explicit ratings or implicit interactions (e.g., likes, clicks, shares, and saves) to learn user preferences and item characteristics. Beyond numerical ratings, textual reviews provide insights into users’ fine-grained preferences and item features. Analyzing these reviews is crucial for enhancing the performance and explainability of personalized recommendation results. In this paper, we provide a comprehensive overview of the development in review-based recommender systems over recent years, highlighting the importance of reviews in recommender systems, as well as the challenges associated with extracting features from reviews and integrating them into ratings. Specifically, we introduce a classification scheme in terms of both the integration of reviews into recommendation systems and the technical methodology. Additionally, we summarize the state-of-the-art methods, analyzing their unique features, effectiveness, and limitations. The study also presents the various evaluation metrics, comparative analysis, datasets, and real-world applications of review-based recommendation systems. Finally, we propose potential directions for future research, including multi-modal data integration, multi-criteria rating information, and ethical considerations.

This paper describes Toronto Metropolitan University's participation in the TREC Clinical Trials Track for 2023. As part of the tasks, we utilize advanced natural language processing techniques and neural language models in our experiments to retrieve the most relevant clinical trials. We illustrate the overall methodology, experimental settings, and results of our implementation for the run submission as part of Team - V-TorontoMU.

Reproducibility underpins scientific progress, yet in recommendation systems (RS) research, it remains one of the most persistent and underappreciated problems. Algorithms have advanced rapidly, but reported performance gains are often fragile and difficult to verify independently. To address this gap, using the PRISMA framework, we systematically review peer-reviewed studies published between 2013, when reproducibility began receiving attention in the RS community, and 2026. We analyze all identified papers and introduce a six-category hierarchical taxonomy of reproducibility challenges in RS. Building on this, we synthesize the solutions proposed in prior work into a structured framework that maps each category of solutions to the challenges it addresses. Our review shows that reproducibility in RS is a property of the entire experimental workflow, not simply the availability of code and data. To this end, we propose an end-to-end Reproducibility Pipeline, spanning data construction, splitting, candidate generation, configuration, training or execution, evaluation, and reporting. Recent advances in LLM-based and agentic recommendation systems introduce new reproducibility risks, including prompt sensitivity, nondeterministic inference, proprietary APIs, multi-step reasoning, tool use, and external dependencies. Finally, we outline future directions for reproducibility in RS. Relevant materials are available at Github.

Background: People affected by cancer often face difficulty finding relevant clinical, psychological, and practical support services. AI-powered navigation assistants may improve access to these resources, but their retrieval performance must be reliable. Objective: To develop a single-pass retrieval-augmented generation assistant for cancer-care navigation and compare the retrieval strategies, including their robustness to reworded questions. Methods: We created a database of 853 cancer-support resources reviewed by librarians, clinicians, researchers, and patient partners. We evaluated the system using 100 questions derived from questions submitted by patients. We compared keyword-based, semantic, and hybrid retrieval using Precision@K, Hit@K, and nDCG@K. The best-performing configuration was then tested using semantically equivalent rewordings of the original questions. Results: Keyword-based retrieval performed poorly, achieving a P@1 of 25.0% and Hit@5 of 43.0%. Semantic retrieval improved these results to 58.0% and 86.0%, respectively. The best hybrid configuration achieved a P@1 of 64.0%, Hit@5 of 90.0%, and nDCG@5 of 51.0%. Performance remained relatively stable when the questions were reworded, with a P@1 of 61.0%, Hit@5 of 88.0%, and nDCG@5 of 46.1%. Conclusions: Hybrid retrieval performed best and remained relatively stable when questions were reworded. However, its limited ability to rank a relevant resource first highlights the limitations of single-pass retrieval for patient-facing cancer navigation. Future work will explore metadata filtering and a multi-agent architecture to improve retrieval reliability.