I reported a recall of 0.699. It was 0.641. Nothing had been miscalculated, and by the time I found it the figure had already reached the slides.
I build knowledge graphs and the machine learning that runs on top of them: ontology design, entity resolution across noisy sources, relationship extraction from unstructured documents, and graph models for link prediction and ranking.
My PhD introduced Inverse Node Frequency, a method for correcting degree bias in path-based link prediction, so that predictions stop concentrating on the entities that are already well understood. I am currently sole engineer on a document-to-knowledge-graph platform for the Koslicki Lab at Pennsylvania State University, and I hold a postdoctoral position at Evotec. Before research I spent four years in investment banking and corporate finance.
I write here mostly about evaluation: how measurements on these systems go wrong, and what stops it happening twice.
Writing
A wrong number is usually a scope error, not an arithmetic one
Elsewhere