Semi-supervised Nonnegative Matrix Factorization for Document Classification

TL;DR

This paper introduces semi-supervised nonnegative matrix factorization models that jointly perform topic modeling and document classification, providing interpretable results and flexible application to various supervised learning tasks.

Contribution

The paper presents new SSNMF models with multiplicative update training methods, combining topic modeling and classification in a unified, interpretable framework.

Findings

Effective on 20 Newsgroups dataset

Applicable to multi-label classification

Flexible for other supervised tasks

Abstract

We propose new semi-supervised nonnegative matrix factorization (SSNMF) models for document classification and provide motivation for these models as maximum likelihood estimators. The proposed SSNMF models simultaneously provide both a topic model and a model for classification, thereby offering highly interpretable classification results. We derive training methods using multiplicative updates for each new model, and demonstrate the application of these models to single-label and multi-label document classification, although the models are flexible to other supervised learning tasks such as regression. We illustrate the promise of these models and training methods on document classification datasets (e.g., 20 Newsgroups, Reuters).