Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Optimizing Modality Weights in Topic Models of Transactional Data

Optimizing Modality Weights in Topic Models of Transactional Data Modern natural language processing models such as transformers operate multimodaldata. In the present paper, multimodal data is explored using multimodal topic modeling ontransactional data of bank corporate clients. A definition of the importance of modality for themodel is proposed on the basis of which improvements are considered for two modeling scenarios:preserving the maximum amount of information by balancing modalities and automatic selectionof modality weights to optimize auxiliary criteria based on topic representations of documents.A model is proposed for adding numerical data to topic models in the form ofmodalities: each topic is assigned a normal distribution with learning parameters. Significantimprovements are demonstrated in comparison with standard topic models on the problem ofmodeling bank corporate clients. Based on the topic representations of the bank’s customers, a90-day delay on the loan is predicted. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Automation and Remote Control Springer Journals

Optimizing Modality Weights in Topic Models of Transactional Data

Download PDF
15 pages

Loading next page...
 
/lp/springer-journals/optimizing-modality-weights-in-topic-models-of-transactional-data-gwvUf0oEy7

References (14)

Publisher
Springer Journals
Copyright
Copyright © Pleiades Publishing, Ltd. 2022
ISSN
0005-1179
eISSN
1608-3032
DOI
10.1134/s00051179220120050
Publisher site
See Article on Publisher Site

Abstract

Modern natural language processing models such as transformers operate multimodaldata. In the present paper, multimodal data is explored using multimodal topic modeling ontransactional data of bank corporate clients. A definition of the importance of modality for themodel is proposed on the basis of which improvements are considered for two modeling scenarios:preserving the maximum amount of information by balancing modalities and automatic selectionof modality weights to optimize auxiliary criteria based on topic representations of documents.A model is proposed for adding numerical data to topic models in the form ofmodalities: each topic is assigned a normal distribution with learning parameters. Significantimprovements are demonstrated in comparison with standard topic models on the problem ofmodeling bank corporate clients. Based on the topic representations of the bank’s customers, a90-day delay on the loan is predicted.

Journal

Automation and Remote Control – Springer Journals

Published: Dec 1, 2022

Keywords: multimodal topic modeling; transactional data; classification; loan delinquency forecast

There are no references for this article.