Keywords: Aggregation Bias, Deployment Bias, Evaluation Bias, Learning Bias, Natural Language Processing Technology, Slang
Author: Future Analytica
Publisher: Medium
Publication Year: 2022
Summary: The following article discusses data ethics in context of natural language processing. Several types of bias that we need to watch out for are learning bias, evaluation bias, aggregation bias, and deployment bias. One big thing to consider is the usage of slang and how slang evolves. Data scientists might need to constantly update resources for natural language processing in order to prevent artificial intelligence from learning hateful speech.