Attachments
Methodology
This report began as a combination product review and experimental research paper on how organizations can build a humanitarian chatbot. The scope of the report changed due to a limited amount of field data available and time constraints. The objectives were modified to address the questions: (1) what is a chatbot; (2) what purpose can it serve in the humanitarian sector; (3) what are the key design concepts; (4) what humanitarian chatbots currently exist; and (5) what are the shortfalls. The report uses elements of a narrative literature review. It is not exhaustive in nature but synthesizes information gathered across sources. Searches were conducted on electronic databases and organizational websites, using authoritative texts and article reference lists. The searches used several select terms and excluded articles focusing in depth on the technical aspects of machine learning, natural language processing and development.
The specificity of this topic also required the inclusion of sources outside of peer-reviewed papers, journals and academic texts. It is acknowledged that sources from corporate websites, news and media outlets and independent authors may exhibit bias.The report is limited in addressing the long-term benefits of chatbots in the humanitarian sector. A detailed product review or case study is recommended as well as an investigation into the lifecycle of a humanitarian chatbot.