Data Skeptic

Hannes Rosenbusch, a University of Amsterdam researcher and fiction author, explores how reader differences often outweigh book differences. He talks about why star ratings mislead, how metadata and reviews fall short, and how LLMs and the ISAAC method can help model individual reading preferences. He also contrasts LLMs as research tools versus creative editors.

Erwin Dervishai, a PhD student at the University of Copenhagen who studies representation learning and recommender systems. He explores what disentanglement means for learned embeddings. He discusses methods for interpreting embeddings, reproducibility challenges, trade-offs between interpretability and accuracy, using metadata and LLMs for denoising, and practical ideas for user control.

Ekaterina (Kat) Filadova, a PhD student in MIT EECS studying strategic learning and algorithmic game theory. She explores how users can coordinate to influence recommender systems. Topics include framing recommenders as multi‑agent games, collective altruism that helps underrepresented content, challenges in distinguishing coordinated behavior from bots, and empirical tests showing collectives can improve minority items.

Anas Buhayh, a CU Boulder information science grad student who builds the SMORES simulator, discusses multi‑stakeholder fairness in recommender systems. He covers algorithm pluralism, mainstream versus niche recommenders, user choice and switching behavior. He also explores tradeoffs for providers and consumers, risks like filter bubbles, and the idea of algorithm stores and reproducible simulations.

Roan Schellingerhout, a fourth-year PhD student at Maastricht University studying explainable multi-stakeholder recommender systems. He talks about explainable job matching, comparing textual, bar chart, and graph explanations. The “healthy friction” study testing real versus random explanations and how people use explanations as info rather than decision rules. Also building knowledge graphs, LLMs for friendly explanations, and plans for fairness and real-world tests.

David Liu, assistant research professor at Cornell focused on fairness in ML and recommenders, discusses PCA-based recommenders and why they can ignore niche and minority users. He introduces power niche users and explains how PCA over-specializes while proposing item-weighted PCA and upweighting strategies. The chat covers tradeoffs, evaluation, scalability, and the need for better datasets.

Cory Zechmann, a seasoned content curator and the mind behind the music blog Silence No Good, delves into the fascinating blend of human curation and machine learning in content discovery. He discusses the cold start problem and the importance of editorial signals in algorithmic systems. Cory emphasizes the role of human curators in enhancing data and mitigating filter bubbles. He also highlights the significance of balancing familiarity with surprise, and the necessity for better metrics to improve personalization. Lastly, he shares insights on how conversational AI might redefine user preferences in the future.

In this discussion, guest Santiago De Leon Martinez, a doctoral researcher at the Kempelin Institute, dives into the innovative use of eye tracking in recommender systems. He reveals the mechanics behind gaze data, fixations, and saccades, showcasing the RecGaze dataset tailored for studying browsing patterns. Santiago highlights how eye tracking can uncover insights beyond traditional click data, addressing positional bias and user engagement. He also addresses ethical concerns and shares his vision for improving recommendation algorithms by simulating user behavior.

Boya Xu, an Assistant Professor of Marketing at Virginia Tech, explores the intricacies of recommender systems. She delves into hybrid approaches that combine collaborative filtering and bandit learning to tackle challenges like the cold start problem for new users. Boya emphasizes using demographic information for bootstrapping recommendations and ensuring fairness for minority users. She also discusses how recommender systems affect consumer behavior and content creation across digital platforms, shedding light on the impact of algorithms in shaping user experiences.

Florian Atzenhofer-Baumgartner is a PhD student at Graz University of Technology, specializing in recommender systems for digital humanities projects like Monasterium.net. He discusses why traditional recommenders fail in complex digital archives, addressing the diverse needs of users from historians to genealogists. Florian elaborates on technical challenges such as sparse interaction matrices and multi-modal similarity approaches. The conversation also highlights the importance of balancing serendipity and utility in recommendations and the unique evaluation metrics for non-commercial domains.