Machine Learning for Humanitarian Data: Tag Prediction using the HXL Standard

Advances in natural language processing and machine learning (ML)enable the automation of humanitarian tasks that would traditionally require the expertise of a human expert.The labor-intensive process of data labeling requires crisis responders to spend valuable hours wrangling data instead of assisting with relief efforts. We present a machine learning model to predict tags for datasets from the United Nations Office for the Coordination of Humanitarian Affairs (UNOCHA) with the labels and attributes of the Humanitarian Exchange Language(HXL) Standard. This paper details the methodology used to predict the corresponding tags and attributes for a given data set with an accuracy of 94% for HXL header tags and an accuracy of 92% for descriptive attributes. Compared to previous work, our workflow provides a 14% accuracy increase and is a novel case study of using ML to enhance humanitarian data.