Grammarly’s Approach: Building Sequence Tagging Approach

Have you ever wondered what the AI behind Grammarly, and its $13B valuation, is all about?

Grammarly applies “best-in-class” NLP and ML methods to allow the platform to provide its users real-time suggestions, offer contextual grammatical error correction, ensure consistency, encourage inclusive language practices, promote conciseness, among other features.

We are delighted to have the authors of the paper “Grammatical Error Correction: Tag, Not Rewrite” at the campus to share their experiences applying the sequence tagging approach to the grammatical error correction task.

Kostiantyn Omelianchuk and Oleksandr Skurzhanskyi are Applied Research Scientists at Grammarly, and in their talk they will present what worked and what didn’t, wins and challenges they encountered along the road, and their future plans in this area. The team will also describe how this approach could be extended to other monolingual tasks, like text simplification.

Don’t miss out on this talk to explore Grammarly’s state-of-the-art approach and to get to know the Grammarly team.

‍

About the speakers: 

Kostiantyn Omelianchuk is a technical lead and natural language processing expert with more than six years of experience in the industry. At Grammarly, Kostia leads a cross-functional team that focuses on improving the quality of grammatical error correction above the level of a proficient human editor. 

Oleksandr Skurzhanskyi leads a project at Grammarly that has helped grow the capabilities of its writing assistance to go way beyond grammar. He has six years of experience in the machine learning field and is interested in such topics as self-supervised learning, non-autoregressive text generation, and multimodal learning.

‍