The Moral Foundations Theory defines foundations to explain human moral reasoning and its role in the decision-making process, including how information is perceived and interpreted.
The Moral Foundations Dictionary (MFD) is a dictionary used to explore variations in moral concerns, based on the Moral Foundations Theory. It was created by Jesse Graham and Jonathan Haidt for use with the Linguistic Inquiry and Word Count (LIWC) program (www.liwc.net).
MFD-BR is the version of MFD in Brazilian Portuguese and can be used to explore variations in moral concerns in Brazilian Portuguese. MFD-BR is designed to be configured with the LIWC program as a customized dictionary.
Like the original MFD, MFD-BR has eleven categories: a general one for morality aspects, five categories for "virtues", and five for "vices". The percentage of words on such categories brings linguistic information about the five foundations for moral judgment.
The development of this version in Brazilian Portuguese is still in its initial stages, but the results are encouraging for further studies. Considering the importance of the theme, studies using this resource can be carried out in a range of areas of interest, such as political and social sciences.
When using this dictionary for academic purposes, please cite our article:
@INPROCEEDINGS{
AUTHOR="Flavio Carvalho and Helder Yukio Okuno and Lais Baroni and Gustavo Guedes",
TITLE="A Brazilian Portuguese Moral Foundations Dictionary for Fake News classification",
BOOKTITLE="39th International Conference of the Chilean Computer Science Society (SCCC 2020)",
ADDRESS="Universidad Católica del Norte, Campus Guayacán, Coquimbo – Chile",
DAYS="16-20",
MONTH="nov",
YEAR="2020",
ABSTRACT="The Moral Foundations Theory defines foundations to explain human moral reasoning and its role in the decision-making process, including how information is perceived and interpreted. A problem related to aspects of moral values that is currently gaining notoriety is the spread of false information known as "Fake News". Natural language processing techniques are being used in social sciences studies to deal with the Fake News detection task. This work introduces and brings details from the development of MFD-BR, a Brazilian Portuguese lexicon based on the Moral Foundations Theory, designed to measure Moral Sentiment in texts. It also contributes to Fake News detection strategies by assessing the difference in moral dimensions to distinguish between reliable sources texts and texts originated from low-reputation sources (considered by fact-checking agencies).",
}