Machine Learning or better deep machine learning has recently popped up again in media mainstream.
In March 2016, Google “AlphaGo” victory against one of the best GO players in the world made headlines. Now, this was not the first time that a machine beat humans in a game. Five years ago, IBM “Watson” project won against former Jeopardy winners.
Well, why the recent hype? Next to mastering the complex game of GO, one aspect could be that all major IT companies (Amazon, Microsoft, Google, Facebook, Baidu, etc) are investing in technology which can replace human decision making with machine decision making based on algorithms.
That also includes SAP which currently pilots 5 applications using machine learning. All were developed by the SAP Innovation Center Network which is SAP’s research organization. The most applicable ones for Public Sector administration are probably “Invoice Matching” and “Curriculum Vitae (CV) Matching”:
Invoice Matching - Matching an incoming payment to an invoice is quite often labor intensive and given the nature of certain government agencies. It even gets more difficult as some people consider the payment of speed tickets or parking ticket unfair and consequently, do not not enter the correct reference figures or fine amount.
CV matching aims to reduce the time-consuming shifting through all CV to come up with a ranked short list of the best candidates for a given open position.
A minimum prerequisite for applicability for machine learning is the existence of critical number of historic decisions. Rare events like earthquakes,floods or terrorist attacks which in most countries happen only occasionally will not necessarily benefit from such a technological advance.
What other areas in a Public sector government administration could benefit from automated decisions and help streamlining business processes at a Public Sector administration:
- Automatically suggesting the right budget address/account assignment combined with a check whether a suitable address has available budget
- Improve the dunning process to increase efficiency for collection
- Check outgoing payment transactions for fraud patterns
- Automate tax return assessments
- Automate eligibility decision in case of grants or social services
Feel free to share further ideas.
Albrecht Weiss
SAP Solution Management Public Sector