Enhancing the SET Based Data Modeling Method with Context Meta Descriptors

摘要

Contextual processing is a new emerging field based on the notion that information surrounding an event lends new meaning to the interpretation of the event. Data mining is the process of looking for patterns of knowledge embedded in a data set. The process of mining data starts with the selection of a data set. This process is often imprecise in its methods as it is difficult to know if a data set for training purposes is truly a high quality representation of the thematic event it represents. Contextual dimensions by their nature have a particularly germane relation to quality attributes about sets of data used for data mining. This paper reviews the basics of the contextual knowledge domain and then proposes a method by which context and data mining quality factors could be merged and thus mapped. It then develops a method by which the relationships among mapped contextual quality dimensions can be empirically evaluated for similarity. Finally, the developed similarity model is utilized to propose the creation of contextually based taxonomic trees. Such trees can be utilized to classify data sets utilized for data mining based on contextual quality thus enhancing data mining analysis methods and accuracy.