Oluwatosin Adewumi

About me

Tosin Adewumi is a researcher working in the intersection of natural/technical language processing (NLP/TLP) and battery lifecycle, with involvement in RECAT, led by Martin Karlsson. I am a member of WASP faculty and do education and ethics (or gender) research as it relates to large language models (LLMs). I am also part of the DAAD AInet Fellows and Alumni Network (German Academic Exchange Service). I successfully defended my Ph.D. in June 2022 under the supervision of Marcus Liwicki as principal supervisor and Foteini Liwicki as assistant supervisor. My Ph.D. thesis is titled "Vector Representations of Idioms in Data-Driven Chatbots"  and it involves the creation of an idioms dataset in English (with over 1,100 cases spread across 10 classes) and training state-of-the-art (SotA) open-domain models, in order to have idiom-aware conversational systems. We also created the Swedish analogy test set for word embeddings and the AfriWOZ dataset for 6 African languages.

My previous position was the Director of Operations at Ephphatha Consulting, in Nigeria. My experience as a professional in the IT industry has spanned both public and private organisations. I am also a member of Nigeria Computer Society (NCS). In 2016, I had the privilege of being the seminar speaker on e-petitions for members of the public petitions committee of the Federal House of Representatives of Nigeria.

Publications of Tosin Adewumi (Google Scholar)