|ru-RU: Антон Белый / zh-CN: 安东·别里 / IPA: [ɐn̪’t̪o̞n̪ ‘bʲeɫɨç]
firstname.lastname@example.org / Scholar / Github / LinkedIn
natural language processing / machine learning / algorithms
I develop efficient algorithms for large-scale information retrieval and structured, semantically-aware text generation. I also design interfaces that allow humans and machines collaborate on generating structured objects such as scripts or schemas.
My background is in Algorithms, Machine Learning and Software Development: prior to PhD, I interned at JetBrains, worked for a year as a Backend Developer at VK and for two years as a Senior Data Scientist at Tochka Bank. I received B.Sc. in Applied Maths and Computer Science from ITMO University, where I’ve collaborated with Andrey Filchenkov and Konstantin Vorontsov on building exploratory search engines and plagiarism detection tools.
|InFillmore: Frame-Guided Language Generation with Bidirectional Context.
Jiefu Ou, Nathaniel Weir, Anton Belyy, Felix Yu, and Benjamin Van Durme. Accepted to *SEM 2021.
|Script Induction As Association Rule Mining.
Anton Belyy and Benjamin Van Durme. ACL Workshop on Narrative Understanding, Storylines, and Events 2020.
|[paper] [slides] [code]|
|Improved Evaluation Framework for Complex Plagiarism Detection.
Anton Belyy, Marina Dubova, and Dmitry Nekrasov. ACL 2018.
|[paper] [poster] [code]|
|Framework for Russian Plagiarism Detection Using Sentence Embedding Similarity and Negative Sampling.
Anton Belyy and Marina Dubova. Dialogue 2018.
|[paper] [slides] [code]|
|Quality Evaluation and Improvement for Hierarchical Topic Modeling.
Anton Belyy, Mariia Seleznova, Alexey Sholokhov, and Konstantin Vorontsov. Dialogue 2018.
|[paper] [slides] [demo]|
My web-dev projects, written (mostly) on the rare occasions of having spare time, and mostly to play with a fancy new Web framework :)
|SchemaBlocks (2019 – present)
A web-based UI for “programming” KAIROS schemas, specifying events, relations, and entity types. Based on Google Blockly.
|Rysearch (2017 – 2019)
An exploratory search engine, that uses topic models to organize popular science literature into a hierarchical map and perform inexact document queries over this map. Based on FoamTree and hARTM.
A karaoke music player, featuring songs based on lyrics of a great 20th-century Russian poet Sergey Esenin. Written overnight from scratch in CoffeeScript.
An online strategy game, where you can build castles, mine resources, wage wars and trade with neighbors… by sending MongoDB database queries! Inspired by mysqlgame.