Jan-Christoph Kalo

Hi, I am Jan-Christoph Kalo, Assistant Professor at the University of Amsterdam in the Intelligent Data Engineering Lab (INDElab).

My research studies what happens at the boundaries between different representations of knowledge. The same fact can live as a sentence in text, a row in a statistical table, a triple in a knowledge graph, a SQL query, a logical formula, or somewhere in the parameters of a language model — and moving between these forms is never lossless, nor is it context-free. Whether two representations actually encode the same knowledge depends on what population, time frame, or scope is assumed, and integrating them means making those assumptions explicit. I work at the intersection of language models, knowledge graphs, and databases on the translations, integrations, and mismatches between these representations, combining semantic web, database, and NLP techniques.

Concretely, this spans information extraction and knowledge base construction, text-to-SQL over real-world statistical data (in collaboration with Statistics Netherlands / CBS), autoformalization and deductive reasoning with LLMs, cross-lingual consistency on large public knowledge sources like Wikipedia and Wikidata, and probing what language models actually represent. A recurring methodological stance runs through these: expressing the same knowledge in multiple forms and studying where the versions disagree is a diagnostic — it tells us what each representation captures, what it silently drops, and where context is doing hidden work.

Beyond research, I supervise MSc and PhD projects and teach databases and knowledge technologies in the Bachelor and Master programs at UvA.