cs.CLSep 18, 2026

CultureMINE: Datasets and Methods for Improving the Cultural Capabilities of NLP Systems

Authors: Tania Chakraborty, Eylon Caplan, Zhaoqing Wu, Kevin Cushing, Han Qin, Shreya Havaldar, Dan Goldwasser

Organizations: Purdue University · University of Pennsylvania

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

In recent years, there has been a surge of interest in Cultural NLP, with substantial efforts to create globally inclusive NLP systems. The rapid growth of literature in this field makes it difficult to track trends in methods and data resources. To address this, we analyze over 375 papers to answer three complementary questions: (1) What Cultural Capabilities (CCs) are being targeted in NLP systems? (2) How are cultural data resources being created? and (3) What methods are being used to improve the CCs of those systems? We discuss trends observed across the three questions, and identify relevant research gaps. To facilitate further research in this field, we release our full list of analyzed papers in the form of an interactive web interface, which includes a feature to allow researchers to add their work; we hope this facilitates future research and proves to be a valuable resource for the Cultural NLP community.

Figures & tables

Figure 1: An overview of our surveying pipeline. We conduct iterative reading and manual annotation of over 375 papers, resulting in (1) our categorization of Cultural Capabilities, (2) new cultural Data and System taxonomies, and (3) key insights for the field. All annotated papers and findings are synthesized into CultureMine , an interactive web repository for the community.
Figure 2: How are Cultural Datasets being created? Our taxonomy for Dataset Creation methods (solid color boxes), and example papers employing these methods (striped color boxes). Note that a paper can combine any number of data creation methods.
Figure 3: What methods are used to improve Cultural Capabilities? Our taxonomy of Systems (solid color boxes), and example papers employing these methods (striped color boxes). Note that a paper may combine any number of methods.
Figure 4: CultureMine interface has dynamic visualizations in the top section. The fields in the bottom, which include our taxonomies for datasets and systems, can be used as filters to find relevant papers. The filters also include information such as cultural proxies used, what cultures, and what languages!
Figure 5: Frequency of future-work clusters in the papers
Appendix figures & tables13 assets

Supplementary material from the paper’s appendix.

Appendix

Figure 6: Filtered Paper List in CultureMine
Figure 7: Cultural Capabilities Filter in CultureMine
Figure 8: Cultural Groups Filter in CultureMine
Figure 9: Method distribution across cultural capabilities
Figure 10: Cultural capability distribution across methods
Figure 11: Dataset creation method combinations stacked by cultural capability
Figure 12: Heatmap of cultural capability by dataset creation method
Figure 13: Pie Chart of Cultural Proxies Used
Figure 14: Bar Chart of the Frequencies of the 10 Most Commonly Studied Languages
PaperCulture capability
Wang et al. (2024a)CK, CT, VCA
Singh et al. (2025c)CK, VCA
Khanuja et al. (2020)SC
Joshi et al. (2025)CSHR
Nguyen et al. (2024a)CCR, SC
Tao et al. (2024)SCA
Appendix
Table 15
PaperCulture capability
Sheng et al. (2016)MCK
Wilson et al. (2016)SC, VCA
Friedman et al. (2019)CSHR, CCR
Maji et al. (2025b)MCK
Cuevas et al. (2025)CSHR
Sahoo et al. (2025)CK
Appendix
Table 16
PaperCulture capability
Fang et al. (2025)CK
Yu et al. (2025)SC
Liu et al. (2025e)MCK, CK
Zhang et al. (2025b)MCK
Togmanov et al. (2025)CK
Yari and Koto (2025)CK
Appendix
Table 17
PaperCulture capability
Malik et al. (2025)SC, VCA
Caplan et al. (2025)SC
Mohammadi et al. (2025)SCA
Bennie et al. (2025)CSHR
Rachamalla et al. (2025)CK, CSHR, SC
Liao et al. (2025)CT
Appendix
Table 18
More in this topic

Cultural Knowledge in Language Models

Cultural knowledge in language models concerns how systems acquire, access, and apply culture-specific facts, norms, meanings, and practices.

See all 127 papers

Explore similar work

CardsList

  1. The Culture Funnel: You Can't Align What isn't in the Data

    Jun 11, 2026Ananya Sahu, Mehrnaz Mofakhami, Daniel D'Souza +3Cultural AlignmentCultural Bias in Language Models

    CohereCorrespondingLabs

  2. MSQA: A Natively Sourced Multilingual and Multicultural SimpleQA Benchmark

    Jul 1, 2026Xianru Chen, Yukai Huang, Mingxiang Chen +6Multilingual Language Model EvaluationLLM Evaluation

    ByteDance Seed · Nanjing University · M-A-P

  3. Toward Culturally Grounded Natural Language Processing

    Mar 27, 2026Sina Bagheri NezhadCross-Cultural Language Model EvaluationCultural Knowledge in Language Models

    Independent Researcher, Seattle, WA, USA