The Investitionsbank Berlin (IBB) funds the project “Future Finance.” In cooperation with the Candis GmbH, the DAI-Lab aims to introduce Artificial Intelligence into accounting and taxation. In particular, small and medium-sized companies as well as freelance professionals have to invest considerable efforts to administer invoices. They would benefit could they spend this time focusing on their core business instead. Candis offers a software package which handles invoices in a semi-automatic fashion. Candis strives to grow and needs to account for economies of scale. These demand a higher degree of automation which must not negatively affect the service quality.
Mapping invoices to transactions represents a central task. We apply machine learning to recognize patters to boost the mapping accuracy. Invoices represent a largely unexplored type of data. As a result, we have to adjust existing approaches and innovate new ones. Rule-based approaches fall short of the dynamics of business communication. Instead, the system has to adapt to changing conditions.