Exclusive Topic Model

Hao Lei (), Kailiang Liu () and Ying Chen ()
Additional contact information
Hao Lei: National University of Singapore
Kailiang Liu: National University of Singapore
Ying Chen: National University of Singapore

Chapter Chapter 3 in Research Papers in Statistical Inference for Time Series and Related Models, 2023, pp 83-109 from Springer

Abstract: Abstract Digital documents are generated, disseminated, and disclosed in books, research papers, newspapers, online feedback, and other content containing large amounts of information, for which discovering topics becomes important but challenging. In the current field of topic modeling, there are limited techniques available to deal with (1) the predominance of frequently occurring words in the estimated topics and (2) the overlap of commonly used words in different topics. We propose exclusive topic modeling (ETM) to identify field-specific keywords that typically occur less frequently but are important for representing certain topics, and to provide well-structured exclusive terms for topics within each topic. Specifically, we impose a weighted LASSO penalty to automatically reduce the dominance of frequently occurring but less relevant words and a pairwise Kullback–Leibler divergence penalty to achieve topic separation. Numerical studies show that the ETM can detect field-specific keywords and provide exclusive topics, which is more meaningful for interpretation and topic detection than other models such as the latent Dirichlet allocation model.

Date: 2023
References: Add references at CitEc
Citations:

There are no downloads for this item, see the EconPapers FAQ for hints about obtaining it.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:spr:sprchp:978-981-99-0803-5_3

Ordering information: This item can be ordered from
http://www.springer.com/9789819908035

DOI: 10.1007/978-981-99-0803-5_3

Access Statistics for this chapter

More chapters in Springer Books from Springer
Bibliographic data for series maintained by Sonal Shukla () and Springer Nature Abstracting and Indexing ().