Stylometry can identify the most likely author of an anonymous text. But which features of the text contribute most to that attribution? My research explores how stylometric methods describe writing style and why they work.
I am a programmer and visiting lecturer at HSE University’s School of Philological Studies, Faculty of Humanities. I develop research software and teach on the Literature and Digital Humanities course.
Together with Evgeny Kazartsev, I work on interpretable stylometric methods. I also contribute to SOCIOLIT and PROZIMETRON and write practical guides to computational text analysis.
My research
My research questions, methods, and findings in a nutshell.
Burrows’s Delta and its variants can group texts by authorship with good accuracy. Yet not all words contribute equally to the result. Let’s decompose Delta into its components to identify the strongest authorial signals.
Genome-wide association studies (GWAS) have proved effective in genomics, revealing statistical links between genetic markers and traits such as height or disease. Let’s apply the same logic to authors’ word-frequency profiles.
Tools for studying Russian literature and its relationship with society.
SOCIOLIT is a project of HSE University’s School of Philological Studies. It brings together texts and tools for exploring word use, comparing corpora, and analysing sentiment through a web interface.
Search for words and lemmas in literary corpora
Compare word distributions across groups of texts
Explore word frequencies and the emotional tone of a text
As a core developer, I work on analytical modules, corpus processing, APIs, and interfaces for researchers.
Articles in System Block
Guides and essays on language, literature, and AI. All articles are in Russian.