A Manifesto for Pattern Bases
Yannis Theodoridis, Michalis Vazirgiannis, Panos Vassiliadis, Barbara Catania, Stefano Rizzi · 2003
A major challenge for the database discipline in the 21 st century is the efficient manipulation of huge volumes of data along with the extraction and management of useful knowledge from these data sets. Nowadays, sophisticated data processing tools, based on data mining, pattern recognition and other knowledge extraction techniques, produce knowledge artifacts that represent large parts of underlying data sets in a concise and meaningful way (e.g., association rules, frequent parts of signals, etc.). In this paper, we will refer to these knowledge artifacts as patterns. We envision pattern repositories, which we call Pattern Base Management Systems (PBMSs), for treating patterns as persistent objects that deserve effective and efficient storing, processing, and querying. Architectural considerations along with functionality requirements for a PBMS are also presented.