Rosina Weber, PhD

Problem

Supervised machine learning algorithms are trained on labeled examples where labels are treated as optimal outputs, implying there is a single best decision. However, cognitive science shows that under uncertainty, limited resources, or time pressure, decision makers rely on cognitive attributes such as risk tolerance that differ across individuals. As a result, people make different decisions in the same situation.

Solution

Capture the risk tolerance reflected in users’ decisions and use it to provide recommendations aligned with their decision patterns, improving trust and reliance.

Benefits

Organizations that use accurate recommender systems can increase customer satisfaction and improve conversion rates.

Publications

• Hu, B., McVay, J., Leung, A., Chan, D., Weber, R. O., de Visser, E., Summerville, A., Ravichandran, B., Zhang, J., Molineaux, M., Ji, H., Basharat, A. (2025, October 10). From Talk to Triage: Pluralism is Necessary but Not Sufficient for AI Alignment. PsyArXiv Preprints https://doi.org/10.31234/osf.io/hdu92_v1. Position Paper.
• Mainali, M., Sureshbabu, H., Sen, A., Rauch, C. B., Meyer, J., Turner, J. T., Floyd, M. W., Molineaux, M., & Weber, R. O. (2025). Classical AI vs. LLMs for decision-maker alignment in health insurance choices . In Proceedings of the Twelfth Annual Conference on Advances in Cognitive Systems (ACS 2025).
• Rauch, C., Molineaux, M., Mainali, M., Sen, A., Floyd, M., & Weber, R. O. (2025). Role-based ethics for decision-maker alignment. Human Alignment in AI Decision-Making Systems (HAADMS). Proceedings of the IEEE Conference on Artificial Intelligence 2025 (IEEE CAI 2025).
• Sen, A., Weber, R., Mainali, M., Rauch, C. B., Turner, J. T., Meyer, J., Floyd, M., Molineaux, M. (2025). Decision Maker Alignment: Benchmark Datasets. Human Alignment in AI Decision-Making Systems (HAADMS). Proceedings of the IEEE Conference on Artificial Intelligence 2025 (IEEE CAI 2025).
• Mainali, M., & Weber, R. O. (2025, April). Exploring Cognitive Attributes in Financial Decision-Making. In METACOG-25: 2nd Workshop on Metacognitive Prediction of AI Behavior, SIAM International Conference on Data Mining (SDM25)
•  Anik Sen, Mallika Mainali, Christopher B. Rauch, Ursula Addison, Michael Floyd, Prateek Goel, Justin Karneeb, Ray Kulhanek, Othalia Larue, David Menager, Matthew Molineaux, JT Turner and Rosina O Weber. Counterfactual-Based Synthetic Case Generation. In Bridge et al. (Eds.): ICCBR 2024, Lecture Notes in Computer Science. Springer.
•  Christopher B Rauch, Ursula Addison, Michael Floyd, Prateek Goel, Justin Karneeb, Ray Kulhanek, Othalia Larue, David Menager, Mallika Mainali, Matthew Molineaux, Adam Pease, Anik Sen, JT Turner, Rosina Weber (2024)Algorithmic Decision-Making in Difficult Scenarios. In AAAI-24 Spring Symposium on Human-Like Learning
•  Rosina O. Weber, Manil Shrestha, and Adam J Johs (2021) Knowledge-based XAI through CBR: There is more to explanations than models can tell. H. Borck, V. Eisenstadt, A. Sanchez-Ruiz, M. Floyd (eds.) Workshops Proceedings for the 29th International Conference on Case-Based Reasoning co-located with the 29th International Conference on Case-Based Reasoning (ICCBR 2021)

Problem

Large language models (LLMs) are the first software that can communicate with humans using natural language. As a result, organizations are seeking to leverage these tools to increase productivity. However, these tools have two main problems: (1) they cannot reason in the classical AI sense, and (2) they do not reliably justify or explain their decisions.

Solution

Use LLM-orchestrated solutions that combine LLMs for natural language tasks with classical AI tools. This combined approach evaluates individual decisions and provides users with the pros and cons of each option.

Benefits

Users make the final decision after reviewing the pros and cons of each alternative, so the selected decision is already explained.

Publications

•  Rauch, C. B., & Weber, R. O. (2026). ProEthica: A Professional Role Based Ethical Analysis Tool Using LLM-Orchestrated, Ontology Supported Case Based Reasoning. Proceedings of the Fortieth AAAI Conference on Artificial Intelligence (AAAI-26).
• Weber, R O, Rauch, C B, Amin, S (2025) Decision Making in LLMs: A First Step. In Martin, K & Ye, X (eds.) 2nd Workshop on Case-Based Reasoning and Large Language Model Synergies (CBR-LLM) at ICCBR2025, June 30, 2025, Biarritz, France. Ceur.

Problem

Evaluating and selecting candidates for recruitment is typically handled by recruiters who may manage thousands of openings per month. Recruiters often use information retrieval tools or large language models to match job descriptions to applicants’ resumes or curriculum vitae (CVs). In practice, these approaches are often inaccurate, excluding many high-quality candidates (low recall) and including many poor candidates (low precision).

Solution

People who are knowledgeable about the job requirements can prepare example resumes that meet the desired criteria, including acceptable variations, such as types of experience that could be considered in place of others. Alternatively, when the acceptable variations of a job description are known, language models can help create example resumes. Using these seed resumes, an algorithm can match applicants’ resumes to the example set.

Benefits

This makes candidate filtering more flexible and better able to capture variations that humans consider acceptable, which are difficult to capture with current approaches. The algorithm can be made more or less strict, adjusting to the number of applicants and the level of demand on a case-by-case basis. The same approach can be used for promotion.

Publications

• Weber, R. O., and Duarte, K.(2021)Data-driven artificial intelligence to automate researcher assessment.Scientometrics, 126(4), 3265-3281. DOI 10.1007/s11192-020-03859-x

•   Duarte, K., Weber, R.O., Pacheco, R.C.S. (2016). Conceptual data model for research collaborators. In CIKI: VI International Conference on Knowledge and Innovation. Bogota’, Colombia. Oct 31- Nov 1^st

•   Duarte, K., Weber, R. O., Pacheco, R.C.S. (2016). Purpose-oriented metrics to assess researcher quality. In proceedings of the 21st International Conference on Science and Technology Indicators (STI2016), pp. 1310-1314

• Duarte, K., Weber, R., Pacheco, R. C. S. (2016). Case-Based Comparison of Career Trajectories.  In: A Coman and S Stelios Kapetanakis (eds): Proceedings of the ICCBR 2016 Workshops, RATIC 2016: Workshop on Reasoning about Time in CBR. Atlanta, Georgia, Oct 30 to Nov 02 (2016)