Topology of Word Embeddings: Singularities Reflect Polysemy

11/18/2020
by   Alexander Jakubowski, et al.
0

The manifold hypothesis suggests that word vectors live on a submanifold within their ambient vector space. We argue that we should, more accurately, expect them to live on a pinched manifold: a singular quotient of a manifold obtained by identifying some of its points. The identified, singular points correspond to polysemous words, i.e. words with multiple meanings. Our point of view suggests that monosemous and polysemous words can be distinguished based on the topology of their neighbourhoods. We present two kinds of empirical evidence to support this point of view: (1) We introduce a topological measure of polysemy based on persistent homology that correlates well with the actual number of meanings of a word. (2) We propose a simple, topologically motivated solution to the SemEval-2010 task on Word Sense Induction Disambiguation that produces competitive results.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
12/06/2019

Recovering the homology of immerged manifolds

Given a sample of an abstract manifold immerged in some Euclidean space,...
research
12/06/2019

Recovering the homology of immersed manifolds

Given a sample of an abstract manifold immersed in some Euclidean space,...
research
02/08/2019

Humor in Word Embeddings: Cockamamie Gobbledegook for Nincompoops

We study humor in Word Embeddings, a popular AI tool that associates eac...
research
10/26/2020

Syllabification of the Divine Comedy

We provide a syllabification algorithm for the Divine Comedy using techn...
research
06/20/2018

The Corpus Replication Task

In the field of Natural Language Processing (NLP), we revisit the well-k...
research
09/23/2021

Putting Words in BERT's Mouth: Navigating Contextualized Vector Spaces with Pseudowords

We present a method for exploring regions around individual points in a ...
research
01/02/2022

On universal sampling representation

For the multivariate trigonometric polynomials we study convolution with...

Please sign up or login with your details

Forgot password? Click here to reset