Aaron Steiner

Researcher

B6, 26, Room C 1.04
68159 Mannheim, Germany

E-Mail: aaron.steineruni-mannheim.de
Phone: +49 (0) 621 / 181-2566

Research Interests

End-to-end data integration with LLMs across schema matching, value normalization, entity matching, and data fusion
Large language models for entity matching: prompting, fine-tuning, and label-efficient training
Web agents and their evaluation, including benchmark design for multi-shop and multi-step reasoning tasks
Agent-environment interfaces: how the choice of interface (MCP, RAG, structured web, raw HTML) shapes agent autonomy and effectiveness

Teaching

Large Language Models and Agents (FSS2026) – Course Details
Data Mining (FSS2026) – Course Details
Seminar CS715: Solving Complex Tasks using Large Language Models (FSS2026) – Course Details
Web Data Integration (HWS2025) – Course Details
Team project: LLM-Agents in Data Integration Pipelines (HWS2025) - Course Details
Large Language Models and Agents (FSS2025) – Course Details
Data Mining (FSS2025) – Course Details
Seminar CS715: Solving Complex Tasks using Large Language Models (FSS2025) – Course Details

Software and Benchmarks

PyDI – Open-source Python framework for end-to-end data integration. Combines traditional string-based methods with embedding- and LLM-based techniques across schema matching, value normalization, entity matching, and data fusion.
WebMall – Multi-shop benchmark for evaluating web agents. 91 tasks across four simulated e-shops with over 4,000 heterogeneous product offers, covering comparison-shopping workflows from product search through checkout.
Billiger.de Products – Bilingual benchmark for evaluating entity matching systems. 13,730 product offers describing 2,168 products across 13 categories, available in German and machine-translated English with identical splits, labels, and identifiers.

Publications

Peeters, R., Steiner, A., Schwarz, L., Caspary, J. Y. and Bizer, C. (2026). WebMall – A multi-shop benchmark for evaluating web agents. In , Proceedings of the 49th International ACM SIGIR Conference on Research and Development in Information Retrieval : SIGIR '26, July 20–24, 2026, Melbourne, Australia (S. 3342-3348). , Association for Computing Machinery: New York, NY.
Steiner, A. and Bizer, C. (2026). Automatic end-to-end data integration using Large Language Models. In , 2026 IEEE 42nd International Conference on Data Engineering Workshops : proceedings, 4–8 May, 2026, Montreal, Canada (S. 142–150). , IEEE: Los Alamitos, CA [u. a.].
Steiner, A., Peeters, R. and Bizer, C. (2026). MCP vs RAG vs NLWeb vs HTML: a comparison of the effectiveness and efficiency of different agent interfaces to the web. In , WWW '26 : Proceedings of the ACM Web Conference 2026 (S. 8493-8496). , Association for Computing Machinery: New York, NY, USA.
Peeters, R., Steiner, A. and Bizer, C. (2025). Entity matching using large language models. In , Proceedings 28th International Conference on Extending Database Technology (EDBT 2025), Barcelona, Spain, March 25-March 28 (S. 529–541). OpenProceedings, OpenProceedings.org: Konstanz.
Steiner, A., Peeters, R. and Bizer, C. (2025). Fine-tuning large language models for entity matching. In , 2025 IEEE 41st International Conference on Data Engineering Workshops : proceedings, 19–23 May 2025 Hong Kong SAR, China (S. 9–17). , IEEE Computer Society: Los Alamitos, CA [u. a.].