Back

February 5, 2019

Data Statements for NLP: Toward Mitigating System Bias and Enabling Better Science

New Research

Experts in information science and computational linguistics propose data statements as a design solution and professional practice for natural language processing technologists to help mitigate issues related to exclusion and bias in Data Statements for NLP: Towards Mitigating System Bias and Enabling Better Science.

In this paper, we propose data statements as a design solution and professional practice for natural language processing technologists, in both research and development. Through the adoption and widespread use of data statements, the field can begin to address critical scientific and ethical issues that result from the use of data from certain populations in the development of technology for other populations. We present a form that data statements can take and explore the implications of adopting them as part of regular practice. We argue that data statements will help alleviate issues related to exclusion and bias in language technology, lead to better precision in claims about how NLP research can generalize and thus better engineering results, protect companies from public embarrassment, and ultimately lead to language technology that meets its users in their own preferred linguistic style and furthermore does not misrepresent them to others.

Read the full paper here. This paper was published in the Transactions of the Association for Computational Linguistics.


Bender, Emily M. and Batya Friedman. 2018. Data Statements for NLP: Toward Mitigating System Bias and Enabling Better ScienceTransactions of the Association for Computational Linguistics 6:587-604.