Creating return on investment for large-scale metadata creation

Abstract

The scholarly communications industry is turning its attention to large-scale metadata creation for enhancing discovery of content. Algorithms used to train Machine Learning are powerful, but need to be used carefully, not least because they can perpetuate bias, racism, and discrimination. Effective use of Machine Learning means facing several technological challenges head-on. This article highlights the specific needs of humanities research to address historical bias and prevent algorithmic bias in creating metadata for Machine Learning. It also argues that the return on investment for large-scale metadata creation begins with building transparency into metadata creation and handling.