Interpretable Topic Extraction and Word Embedding Learning using row-stochastic DEDICOM
Computation and Language
2025-07-23 v1 Artificial Intelligence
Machine Learning
Abstract
The DEDICOM algorithm provides a uniquely interpretable matrix factorization method for symmetric and asymmetric square matrices. We employ a new row-stochastic variation of DEDICOM on the pointwise mutual information matrices of text corpora to identify latent topic clusters within the vocabulary and simultaneously learn interpretable word embeddings. We introduce a method to efficiently train a constrained DEDICOM algorithm and a qualitative evaluation of its topic modeling and word embedding performance.
Cite
@article{arxiv.2507.16695,
title = {Interpretable Topic Extraction and Word Embedding Learning using row-stochastic DEDICOM},
author = {Lars Hillebrand and David Biesner and Christian Bauckhage and Rafet Sifa},
journal= {arXiv preprint arXiv:2507.16695},
year = {2025}
}
Comments
Accepted and published at CD-MAKE 2020, 20 pages, 8 tables, 8 figures