A knowledge-based platform for the classification of accounting documents

Due to the lack of standardization, the task consisting in the classification of invoices’ billing entries for accounting purposes has yet not been made fully automatic despite the diffusion of e-invoicing systems. Each accounting firm adopts its own conventions also on the basis of the specific business domain of the invoice owner. Here, we describe a knowledge-based software platform devised to adequately address this issue. The paper focuses on the evaluation of the most appropriate classification algorithms which are suitable to be employed in an adaptive meta-learning approach. The effectiveness of the selected algorithms is experimentally assessed by running them on a real dataset of invoice entries.