Fairness, Accountability and Transparency while treating gender as a variable.

Scientists and engineers in the machine learning or natural language processing fields should study the ethical implications of the careful design of a variable or a category for reporting results or research work. The natural language processing (NLP) field deals with solving linguistic problems computationally. This paper analyzes the role of accountability held by the NLP engineers and researchers and then propose some solutions as an ethical framework. The theory analyzed here lays down some principles for scientists, engineers and peer reviewers which could be applied on other social variables like race. This paper lays down the foundation of fairness, accountability and transparency principles when treating gender as a variable. The objectives of this paper are:

1. To investigate how gender is perceived from different viewpoints.
2. To define how different the role of scientists and engineers are and who should be accountable for their roles.
3. To critically analyze how gender as a variable was treated in the prior works of Natural Language Processing
4. To present guidelines on how to treat gender as a variable.
5. To extrapolate on how these concerns can be extended to race as a variable.
6. To apply the proposed framework solution.

–> Link to the detailed Report.