Hypothesis testing approach on noisy cases in RICAD

Enabling database applications to perform intelligent record retrieval is one of the most important issues in database research. From one perspective, this particular issue has also been investigated in artificial intelligence (AI) research. Case-based reasoning (CBR) is an approach in AI that focusses on a similar issue. CBR systems mainly try to find the most similar cases from their case bases, and propose their answers based on the found cases. However, the main problem with this approach is that noisy cases can directly affect the accuracy of proposed solutions. This problem can also occur in database applications, if they also intend to formulate the correct answer for their users rather than just retrieving the records. This paper reviews the current practice in CBR research, especially how the CBR systems are dealing with the problem of noisy cases, and describes how the CBR system RICAD deals with noisy cases.