Unsupervised document classification integrating web scraping, one-class SVM and LDA topic modelling

Anton Thielmann; Christoph Weisser; Astrid Krenz; Benjamin Säfken

doi:10.1080/02664763.2021.1919063

Unsupervised document classification integrating web scraping, one-class SVM and LDA topic modelling

J Appl Stat. 2021 Apr 27;50(3):574-591. doi: 10.1080/02664763.2021.1919063. eCollection 2023.

Authors

Anton Thielmann¹, Christoph Weisser^{1

2}, Astrid Krenz^{1

3}, Benjamin Säfken^{1

2}

Affiliations

¹ Center for Statistics, Georg-August-Universität Göttingen, Göttingen, Germany.
² Campus-Institut Data Science (CIDAS), Göttingen, Germany.
³ Digital Futures at Work Research Centre, University of Sussex, Brighton, UK.

Abstract

Unsupervised document classification for imbalanced data sets poses a major challenge. To obtain accurate classification results, training data sets are often created manually by humans which requires expert knowledge, time and money. Depending on the imbalance of the data set, this approach also either requires human labelling of all of the data or it fails to adequately recognize underrepresented categories. We propose an integration of web scraping, one-class Support Vector Machines (SVM) and Latent Dirichlet Allocation (LDA) topic modelling as a multi-step classification rule that circumvents manual labelling. Unsupervised one-class document classification with the integration of out-of-domain training data is achieved and >80% of the target data is correctly classified. The proposed method thus even outperforms common machine learning classifiers and is validated on multiple data sets.

Keywords: LDA topic model; Unsupervised document classification; machine learning; one-class SVM; out-of-domain training data; web scraping.