Abstract: We are surrounded by ubiquitous and interconnected soft-ware systems which gather valuable data. The analysis of these data, although highly relevant for decision making, cannot be performed directly by business users, as its execution requires very specific technical knowledge in areas such as statistics and data mining. One of the complexity problems faced when constructing an analysis of this kind resides in the fact that most data mining tools and techniques work exclusively over tabular-formatted data, preventing business users from analyzing excerpts of a data bundle which have not been previously traduced into this format by an expert. In response, this work presents a set of transformation patterns for automatically generating tabular data from domain models. The described patterns have been integrated into a language, which allows business users to specify the elements of a domain model that should be considered for data analysis.

Authorship

Download reference