User considerations when applying machine learning technology to accounting tasks

Smith, Liezl (2018-12)

Thesis (MCom)--Stellenbosch University, 2018.

Thesis

ENGLISH SUMMARY : Machine learning is a strategic technology that can have an important effect on business, as it is able to perform tasks efficiently that were previously only performed by humans. When implementing this technology in the relevant business processes and utilising it effectively, users have to understand both it as well as other aspects have to be considered. It was found that one area that is well suited to the adoption of machine learning, is accounting. In addition, prior research has shown a need for accounting users to be educated in machine learning as part of their professional training. Therefore, the aim of this study was to enhance users’ understanding of machine learning technology specifically in the performance of accounting processes. A grounded theory methodology was employed to identifying the accounting tasks machine learning could perform, to describe how machine learning functions and to identify the risks, benefits and limitations associated with the technology. Finally, steps and considerations when implementing machine learning technology in the accounting process were provided. The findings of this research are that the user has a key role to play when using machine learning technology in the accounting processes and thus has to understand the technology, the risks and limitations, as well as the benefits of the technology. The risks discussed relate not only to machine learning technology but also to all the components that enable the functioning of the technology to ensure alignment with the accounting process goals. Based on these findings, this research presents the user considerations and steps to take when implementing machine learning in selected accounting processes. These can be used to identify areas that may require attention when a business is adopting machine learning. One important consideration is the implementation of adequate data governance. This is because most of the risks identified for machine learning technology are data risks. Further research could therefore be directed at developing a data governance framework for machine learning technologies.

AFRIKAANSE OPSOMMING : Masjienleer is 'n strategiese tegnologie wat 'n belangrike uitwerking kan hê op besigheid, aangesien dit take doeltreffend kan uitvoer wat voorheen net deur mense uitgevoer is. Wanneer hierdie tegnologie in die toepaslike besigheids prosesse geïmplementeer en doeltreffend benut word, moet gebruikers dit verstaan en verskeie ander aspekte oorweeg. Daar is bevind dat Rekeningkunde een area is wat goed geskik is vir die aanneming van masjienleer. Daarbenewens, het vorige navorsing bevind dat rekeningkundige gebruikers opgelei moet word in masjienleer as deel van hul professionele opleiding. Die doel van hierdie studie was dus om gebruikers se begrip van masjienleertegnologie te verbeter, spesifiek in die uitvoering van rekeningkundige prosesse. 'n Gefundeerde teorie navorsingsmetodologie is gebruik om die rekeningkundige take wat masjienleer kan uitvoer te identifiseer, te beskryf hoe masjienleer funksioneer en om die risiko's, voordele en beperkings wat met die tegnologie verband hou, te identifiseer. Ten slotte is stappe en oorwegings tydens die implementering van masjienleertegnologie in die rekeningkundige proses verskaf. Die bevindinge van hierdie navorsing is dat die gebruiker 'n sleutelrol speel wanneer masjienleertegnologie in die rekeningkundige prosesse gebruik word en dus moet die gebruiker die tegnologie, die risiko's en beperkings, sowel as die voordele van die tegnologie verstaan. Die risiko's wat bespreek word, hou nie net verband met masjienleertegnologie nie, maar ook met al die komponente wat die funksionering van die tegnologie moontlik maak om belyning met die doelwitte van die rekeningkundige proses te verseker. Op grond van hierdie bevindinge, bied hierdie navorsing die gebruikersoorwegings en die stappe om te neem wanneer masjienleer in geselekteerde rekeningkundige prosesse geïmplementeer word. Hierdie oorwegings en stappe kan gebruik word om areas te identifiseer wat aandag benodig wanneer 'n besigheid masjienleer implementeer. Een belangrike oorweging is die implementering van voldoende databeheer, aangesien die meeste van die risiko's wat vir masjienleertegnologie geïdentifiseer is, data-risiko's is. Verdere navorsing kan dus gerig word op die ontwikkeling van 'n data-beheerraamwerk vir masjienleertegnologieë.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/105005
This item appears in the following collections: