Identification of Anomalies in Processes of Database Alteration

Data, especially in large item sets, hide a wealth of information on the processes that have created and modified them. Often, a data-field or a set of data-fields are not modified only through well-defined processes, but also through latent processes; without the knowledge of the second type of processes, testing cannot be considered exhaustive. As a matter of fact, changes in the data deriving from unknown processes can cause anomalies not detectable by testing, which focuses on known data variation rules. History of data variations can yield information about the nature of the changes. In my work I focus on the elicitation of an evolution profile of data: the values data may assume, the change frequencies, the temporal variation of a piece of data in relation to other data, or other constraints that are directly connected to the reference domain. The profile of evolution is then used to detect anomalies in the database state evolution. Detecting anomalies in the database state evolution could strengthen the quality of a system, since a data anomaly could be the signal of a defect in the applications populating the database.