Exploring Dynamic Granules for Time-Varying Big Data

Traditional database management systems (DBMS) offer rich mechanisms and algorithms to manipulate various forms of granules, but mainly for static and crispy data. By applying rough set theory, the Infobright approach shows how granular computing can make important contribution to data management, due to its power of dealing with uncertainty. In particular, we are interested in how to deal with uncertainty in time-varying big data. An examination of the FBeM approach provides useful clue for achieving an abstraction on common features of dynamic data. Other approaches or studies, including behavior mining, deep data and quantum databases also offer useful hindsight for developing a theoretical framework of studying such kind of data. As the result of these examinations, we propose an important concept of dynamic granule, and outline important steps towards the development of a framework involving dynamic granules, for better management of time-varying data at the big data era.