This model is a fine-tuned version of RoBERTa-large [1]. It was trained on 1,400 employee reviews to measure corporate culture. More specifically, it measures the culture dimension “market” of the Competing Values Framework [2,3]. An organization that exhibits a market culture is characterized by an emphasis to compete [2].
The model assigns one of three possible labels:
0 (neutral): Text does not allow any inference about a market culture.
1 (positive): Text contains information in line with a market culture.
2 (negative): Text contains information in opposite to a market culture.
For details on the model and its performance, see Koch and Pasch (2023). Please cite this article when using the model:
S. Koch and S. Pasch, "CultureBERT: Measuring Corporate Culture With Transformer-Based Language Models," in 2023 IEEE International Conference on Big Data (BigData), pp. 3176-3184. doi: 10.1109/BigData59044.2023.10386765
Please see the following
tutorial on how to apply CultureBERT to measure corporate culture in your own text documents:
https://github.com/Stefan-Pasch/CultureBERTOther References:
[1] Liu, Y.; Ott, M.; Goyal, N.; Du, J.; Joshi, M.; Chen, D.; ... & Stoyanov, V. (2019). Roberta: A robustly optimized bert pretraining approach. arXiv preprint arXiv:1907.11692.
[2] Cameron, Kim S.; Quinn, Robert E. (2011): Diagnosing and Changing Organizational Culture. Based on the Competing Values Framework. 3rd ed. San Francisco (CA): Jossey-Bass.
[3] Quinn, Robert E.; Rohrbaugh, John (1983): A Spatial Model of Effectiveness Criteria: Towards a Competing Values Approach to Organizational Analysis. In Management Science 29 (3), pp. 363–377. DOI: 10.1287/mnsc.29.3.363.
