I am currently writing my PhD dissertation, NLP Methods for Automated Fact-Checking — a work in progress.
The dissertation aggregates the evolution of the state of the art in automated fact-checking between the years 2021 and 2026, tracing how the field has shifted from early retrieval-augmented pipelines through neural NLI-based verdict prediction to the current frontier of agentic large language model fact-checking systems.
The text focuses on my own contributions to this progression — including dataset construction (CsFEVER, CTKFacts), system development for shared tasks (AVeriTeC, FEVER 8), and, most recently, work on agentic LLM pipelines that decompose claims, gather evidence autonomously, and produce veracity judgements with natural-language justifications.
The current PDF build is available here and is updated continuously as writing progresses.