|
View: |
Part 1: Document Description
|
|
Citation |
|
|---|---|
|
Title: |
Replication Data for: Cross-lingual classification of political texts using multilingual sentence embeddings |
|
Identification Number: |
doi:10.7910/DVN/OLRTXA |
|
Distributor: |
Harvard Dataverse |
|
Date of Distribution: |
2022-10-04 |
|
Version: |
1 |
|
Bibliographic Citation: |
Licht, Hauke, 2022, "Replication Data for: Cross-lingual classification of political texts using multilingual sentence embeddings", https://doi.org/10.7910/DVN/OLRTXA, Harvard Dataverse, V1, UNF:6:rG8yuayRT3euKCJ2meYa8A== [fileUNF] |
|
Citation |
|
|
Title: |
Replication Data for: Cross-lingual classification of political texts using multilingual sentence embeddings |
|
Identification Number: |
doi:10.7910/DVN/OLRTXA |
|
Authoring Entity: |
Licht, Hauke (University of Cologne, Cologne Center for Comparative Politics) |
|
Distributor: |
Harvard Dataverse |
|
Access Authority: |
Licht, Hauke |
|
Depositor: |
Code Ocean |
|
Holdings Information: |
https://doi.org/10.7910/DVN/OLRTXA |
|
Study Scope |
|
|
Keywords: |
Social Sciences, Social Sciences, multilingual embedding, multilingual text analysis, supervised machine learning |
|
Abstract: |
Established approaches to analyze multilingual text corpora require either a duplication of analysts' efforts or high-quality machine translation (MT). In this paper, I argue that multilingual sentence embedding (MSE) is an attractive alternative approach to language-independent text representation. To support this argument, I evaluate MSE for cross-lingual supervised text classification. Specifically, I assess how reliably MSE-based classifiers detect manifesto sentences' topics and positions compared to classifiers trained using bag-of-words representations of machine-translated texts, and how this depends on the amount of training data. These analyses show that when training data is relatively scarce (e.g. 20K or less labeled sentences), MSE-based classifiers can be more reliable and are at least no less reliable than their MT-based counterparts. Further, I examine how reliable MSE-based classifiers label sentences written in languages not in the training data, focusing on the task of discriminating sentences that discuss the issue of immigration from those that do not. This analysis shows that compared to the within-language classification benchmark, such "cross-lingual transfer" tends to result in fewer reliability losses when relying on the MSE instead of the MT approach. This study thus presents an important addition to the cross-lingual text analysis toolkit. |
|
Methodology and Processing |
|
|
Sources Statement |
|
|
Data Access |
|
|
Other Study Description Materials |
|
|
File Description--f6544545 |
|
|
File: runtimes.tab |
|
|
|
|
Notes: |
UNF:6:rG8yuayRT3euKCJ2meYa8A== |
|
List of Variables: |
|
|
Variables |
|
|
f6544545 Location: |
Variable Format: character Notes: UNF:6:26jKf3hOmLtgwMsvplqvhQ== |
|
f6544545 Location: |
Variable Format: character Notes: UNF:6:hLxDzawegW+9qdAZXSgaGw== |
|
f6544545 Location: |
Variable Format: character Notes: UNF:6:fMBZuxIDrmTKNMShzZKLPg== |
|
f6544545 Location: |
Summary Statistics: Valid 5.0; Min. 0.0172863094011943; Mean 4.627563992222143; StDev 5.181454336286318; Max. 12.2113724167479 Variable Format: numeric Notes: UNF:6:Gktc8PcOafe/RNHh+z7DJQ== |
|
Label: |
analysis1.RData |
|
Notes: |
application/x-rlang-transport |
|
Label: |
analysis2.RData |
|
Notes: |
application/x-rlang-transport |
|
Label: |
baseline.RData |
|
Notes: |
application/x-rlang-transport |
|
Label: |
crosslingual_transfer.RData |
|
Notes: |
application/x-rlang-transport |
|
Label: |
exampleS1.tex |
|
Notes: |
application/x-tex |
|
Label: |
exampleS2.tex |
|
Notes: |
application/x-tex |
|
Label: |
exampleS3.tex |
|
Notes: |
application/x-tex |
|
Label: |
exampleS4.tex |
|
Notes: |
application/x-tex |
|
Label: |
exampleS5.tex |
|
Notes: |
application/x-tex |
|
Label: |
exampleS6.tex |
|
Notes: |
application/x-tex |
|
Label: |
figure1.pdf |
|
Notes: |
application/pdf |
|
Label: |
figure2.pdf |
|
Notes: |
application/pdf |
|
Label: |
figure3.pdf |
|
Notes: |
application/pdf |
|
Label: |
figure4.pdf |
|
Notes: |
application/pdf |
|
Label: |
figureS3.pdf |
|
Notes: |
application/pdf |
|
Label: |
figureS4.pdf |
|
Notes: |
application/pdf |
|
Label: |
figureS5.pdf |
|
Notes: |
application/pdf |
|
Label: |
figureS6.pdf |
|
Notes: |
application/pdf |
|
Label: |
figureS7.pdf |
|
Notes: |
application/pdf |
|
Label: |
figureS8.pdf |
|
Notes: |
application/pdf |
|
Label: |
output |
|
Notes: |
text/plain; charset=US-ASCII |
|
Label: |
results-03843374-51b2-4ddf-b135-020a7b72471b.zip |
|
Text: |
Ported CO capsule |
|
Notes: |
application/zip |
|
Label: |
table1.tex |
|
Notes: |
application/x-tex |
|
Label: |
table2.tex |
|
Notes: |
application/x-tex |
|
Label: |
tableS10.tex |
|
Notes: |
application/x-tex |
|
Label: |
tableS11.tex |
|
Notes: |
application/x-tex |
|
Label: |
tableS12.tex |
|
Notes: |
application/x-tex |
|
Label: |
tableS13.tex |
|
Notes: |
application/x-tex |
|
Label: |
tableS14.tex |
|
Notes: |
application/x-tex |
|
Label: |
tableS15.tex |
|
Notes: |
application/x-tex |
|
Label: |
tableS1.tex |
|
Notes: |
application/x-tex |
|
Label: |
tableS2.tex |
|
Notes: |
application/x-tex |
|
Label: |
tableS3.tex |
|
Notes: |
application/x-tex |
|
Label: |
tableS4.tex |
|
Notes: |
application/x-tex |
|
Label: |
tableS5.tex |
|
Notes: |
application/x-tex |
|
Label: |
tableS6.tex |
|
Notes: |
application/x-tex |
|
Label: |
tableS7.tex |
|
Notes: |
application/x-tex |
|
Label: |
tableS8.tex |
|
Notes: |
application/x-tex |
|
Label: |
tableS9.tex |
|
Notes: |
application/x-tex |