Unsupervised Induction of Linguistic Structures from Text

Note that this invited talk will take place at the American University of Armenia.

Alexander Panchenko will share how graph clustering algorithms can be applied to large lexical networks to induce word sense representation and other structures.

Alexander is an Assistant Professor at Skolkovo Institute of Science and Technology (Skoltech) since 2019 leading a NLP group and a joint MTS-Skoltech laboratory.

We will look at several algorithms: Chinese Whispers, Markov Clustering, and Watset, and discover various applications to linguistic data. Finally, we will briefly discuss how the extracted graph structures can be vectorized using graph embeddings and other related approaches.