A framework for analysis of data freshness

Data freshness has been identified as one of the most important data quality attributes in information systems. This importance increases particularly in the context of distributed systems, composed of a large set of autonomous data sources, where integrating data having different freshness may lead to semantic problems. There are various definitions of data freshness in the literature, depending on the applications where they are used, as well as different metrics to measure them. This paper presents an analysis of these definitions and metrics and proposes a taxonomy based upon the nature of the data, the type of application and the synchronization policies underlying the multi-source information system. We analyze, in terms of the taxonomy, the way freshness is defined and used in several types of systems and we present some open research problems in the field of data freshness evaluation.