Rough Mereology is a paradigm allowing for a synthesis of main ideas of two
potent paradigms for reasoning under uncertainty: Fuzzy Set Theory and Rou
gh Set Theory. Approximate reasoning is based in this paradigm on the predi
cate of being a part to a degree. We present applications of Rough Mereolog
y to the important theoretical idea put forth by Lotfi Zadeh (1996, 1997),
i.e., Granularity of Knowledge: We define granules of knowledge by means of
the operator of mereological class and we extend the idea of a granule ove
r complex objects like decision rules as well as decision algorithms. We ap
ply these notions and methods in the distributed environment discussing com
plex problems of knowledge and granule fusion. We express the mechanism of
complex granule formation by means of a formal grammar called Synthesis Gra
mmar defined over granules of knowledge, granules of classifying rules, or
over granules of classifying algorithms. We finally propose hybrid rough-ne
ural schemes bridging rough and neural computations.(1)