Case-based reasoning (CBR) has found many applications in learning, reasoning, and planning activities. A typical CBR framework attempts to match up the situation at hand with a historical scenario (or case) in its case library. If such a case is found, then the stored solution can be safely applied to the current situation. In almost all real-world applications, it is rare that an old case will exactly match a new situation. At this point, a good CBR architecture tries to find the best matches and generate new test solutions. If the test solution is found to be successful, the case library is updated with a new case—to be used for future reference.
This short paper describes a prediction application: the CBR system is applied to predict when particular payments associated with invoicing are expected. In this type of reasoning, time seems to be used as an explicit parameter in the CBR system. Due to the limited length of the paper, it is not explained if and how the CBR system performs the reasoning in time. Readers will not gain any insight into the architecture, data representation, indexing, and related issues. The authors do present an example of a successful prediction. Readers are referred to a thesis for specific details.