Andrej Risteski

Automated WordNet Construction Using Word Embeddings

We present a fully unsupervised method for automated construction of WordNets based upon recent advances in distributional representations of sentences and word-senses combined with readily available machine translation tools. The approach requires very few linguistic resources and is thus extensible to multiple target languages. To evaluate our method we construct two 600-word test sets for word-to-synset matching in French and Russian using native speakers and evaluate the performance of our method along with several other recent approaches. Our method exceeds the best language-specific and multi-lingual automated WordNets in F-score for both languages. The databases we construct for French and Russian, both languages without large publicly available manually constructed WordNets, will be publicly released along with the test sets.

Skeletal configurations of ribbon trees

The straight skeleton construction creates a straight-line tree from a polygon. Motivated by moduli spaces from algebraic geometry, we consider the inverse problem of constructing a polygon whose straight skeleton is a given tree. We prove there exists only a finite set of planar embeddings of a tree appearing as straight skeletons of convex polygons. The heavy lifting of this result is performed by using an analogous version of Cauchys arm lemma. Computational issues are also considered, uncovering ties to a much older angle bisector problem.