Home // International Journal On Advances in Intelligent Systems, volume 14, numbers 1 and 2, 2021 // View article

Implementing Ethical Issues into the Recommender Systems Design Using the Data Processing Pipeline

Authors:
Olga Levina

Keywords: socio-technical systems; machine learning based systems design; ethical values; ethical analysis; business process

Abstract:
Applying information systems within a business process requires a good understanding of the expected benefits, system requirements as well as of the effects that the process change will have on its actors and stakeholders. Integrating machine learning based systems (MLS) into a business process requires an even broader focus on potentially affected users and stakeholders. Leading to changes in the process, but also in the user and stakeholder behavior, ethical values are directly influenced by the decisions taken during the data processing stages within system development. In this paper, a scenario of an MLS, a fictional recommender system for food delivery, is used to identify potential ethical issues that occur during the composition and usage of the artifact. Data centered analysis of the system development is applied to identify, which ethical values are mostly affected in each data processing stage. It is argued that even when the used data for MLS is not originated from an individual, and thus is not necessarily subject to privacy regulations, ethical analysis and socially-aware engineering of the information system are still required. Suggestions what ethical aspects can be implemented into the design of the MLS are derived here based on the presented scenario. The effects of MLS application in a business process are furthermore briefly outlined for every stage of data processing. Using this scenario-based approach allows identification of social and technical aspects that can be affected by the application of MLS in business context.

Pages: 153 to 163

Copyright: Copyright (c) to authors, 2021. Used with permission.

Publication date: December 31, 2021

Published in: journal

ISSN: 1942-2679