Cross-Corpora Comparisons of Topics and Topic Trends

Victor Bystrov, Naboka Viktoriia (), Anna Staszewska-Bystrova and Peter Winker
Additional contact information
Naboka Viktoriia: Justus Liebig University Giessen, Licher Strasse 64, 35394 Giessen, Germany

Journal of Economics and Statistics (Jahrbuecher fuer Nationaloekonomie und Statistik), 2022, vol. 242, issue 4, 433-469

Abstract: Textual data gained relevance as a novel source of information for applied economic research. When considering longer periods or international comparisons, often different text corpora have to be used and combined for the analysis. A methods pipeline is presented for identifying topics in different corpora, matching these topics across corpora and comparing the resulting time series of topic importance. The relative importance of topics over time in a text corpus is used as an additional indicator in econometric models and for forecasting as well as for identifying changing foci of economic studies. The methods pipeline is illustrated using scientific publications from Poland and Germany in English and German for the period 1984–2020. As methodological contributions, a novel tool for data based model selection, sBIC, is impelemented, and approaches for mapping of topics of different corpora (including different languages) are presented.

Keywords: topic models; text analysis; latent Dirichlet allocation; singular Bayesian information criterion; topic matching (search for similar items in EconPapers)
JEL-codes: C49 (search for similar items in EconPapers)
Date: 2022
References: Add references at CitEc
Citations: View citations in EconPapers (2)

Downloads: (external link)
https://doi.org/10.1515/jbnst-2022-0024 (text/html)

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:jns:jbstat:v:242:y:2022:i:4:p:433-469:n:3

DOI: 10.1515/jbnst-2022-0024

Access Statistics for this article

Journal of Economics and Statistics (Jahrbuecher fuer Nationaloekonomie und Statistik) is currently edited by Peter Winker

More articles in Journal of Economics and Statistics (Jahrbuecher fuer Nationaloekonomie und Statistik) from De Gruyter
Bibliographic data for series maintained by Peter Golla ().