Natural Language Generation Requirements for Social Robots in Sub-Saharan Africa

Abstract

Robots are deployed in Africa mainly in manufacturing, yet they may assist in society as future oriented technologies as well. They may ameliorate, e.g., service delivery issues and skills shortages. In this discussion paper, several uses and use cases relevant to Sub-Saharan Africa are described and requirements identified. We zoom in on human-robot interaction in Niger-Congo B (‘bantu’) languages. Use cases for healthcare and education elucidate specific requirements for the natural language generation component of robots in society. In contrast to typical generation systems, it demands i) combining data-to-text and knowledge-to-text in one system, ii) generating different types of sentences so as to switch between written and spoken language, and iii) processing non-trivial numbers.