CEU Electronic Theses and Dissertations, 2024
| Author | Blanka Toth |
|---|---|
| Title | A Novel Natural Language Processing Framework for Identifying Corporate Greenwashing |
| Summary | This thesis introduces a natural language processing (NLP) based framework employing the “Bidirectional Encoder Representations from Transformers” (BERT) deep learning model to detect greenwashing in corporate communication. This framework, albeit resting on predefined literature, is unique in its approach and methodology. The model utilizes multiple definitions of greenwashing created based on a literature review (deception and misinformation, misleading communication, selective disclosure, and greenwashing as decoupling), to capture the nuanced language of corporate sustainability communication. Validated on a diverse dataset from Google News article titles of the Dow Jones Industrial Average companies, the model was evaluated using precision, recall, F1-score, and Chi-squared test, and its results proved to be statistically significant. The model identified verified cases of greenwashing with a precision of 91.7%, which is considered high, but the lower recall rate of 55% signals that detecting all forms of greenwashing remains challenging. However, the definitions that were created based on an extensive literature review significantly enhanced the model's ability to differentiate and detect different greenwashing approaches, contributing both to theoretical understanding and practical applications. The proposed framework is an improved version of existing greenwashing detection methods that rely on keyword search as a rule of thumb, thereby supporting regulators to consumers by advocating for the more transparent disclosure of corporate environmental practices and environmental impact. Keywords: greenwashing detection, natural language processing, corporate greenwashing, transformers, sustainability reporting |
| Supervisor | Illes, Zoltan |
| Department | Environment Sciences and Policy MSc |
| Full text | https://www.etd.ceu.edu/2024/toth_blanka.pdf |
Visit the CEU Library.
© 2007-2021, Central European University