Supporting Practitioners and Analysts in Making Decisions about Data Analytics to Study Humanitarian Assistance and Disaster Relief Operations
Overview
Using computational methods, such as techniques from AI, machine learning and natural language processing, to study, support, and plan for Humanitarian Assistance and Disaster Relief (HADR) operations, requires practitioners and analysts to make a plethora of decisions. These unavoidable decisions include choices related to sampling and data collection, preparing or pre-processing data for analysis, implementing algorithms, measuring effects, and validating results. What is the impact of these choices on the resulting findings and derived implications, and how can we avoid introducing biases into our findings? In this talk, I address these questions by presenting findings from our work on assessing the impact of choices that analysts and end users of software solutions have to make when collecting and analyzing large-scale text data related to emergency management situations. I highlight best practices for selection data sources, methods and algorithms, explain causes for potential biases, and suggest strategies for mitigating these biases.
Presenter
For more information, see http://jdiesnerlab.ischool.illinois.edu/.