Features of Human-Centred Algorithm Design

Abstract

Algorithms are pervasive, unseen influencers of decisions. Algorithmic features can fluctuate widely, depending on use, user or criteria applied. This paper considers the nascent field of human-centred algorithm design (HCAD), intersecting human-centred design and algorithmic systems. Human-centred, more-than-metric feature selection approaches, create fairer and deeper meaning. More value is created. The unique impact of this paper is to integrate feature selection within a technology HCAD strategy, for a novel, innovative HCAD approach to machine learning. This flexible and evaluative approach can support data advances with human-social nuance, designed for purpose with knowledge for data-driven decisions. The design of machine learning algorithms to the uses in which they will be employed is user-centric. This is important within environments utilising automated, semi-automated or high-performance analytics.