Filter — Trender — Interpretation: A design pattern for interpreting noisy and voluminous data

In this paper we propose and describe a new design pattern called FTI (Filter - Trender - Interpretation) for interpreting noisy and voluminous data sets. FTI consists of 3 consecutive processes: Filter which takes the original data and removes outliers and noise; given large volumes of data, Trender takes the filtered data and abstracts trends; and Interpretation uses rules from knowledge bases to perform qualitative reasoning on the trends to provide an analysis of the original data. In this paper we also show how FTI has been successfully applied to two different case studies.