The DMCS system is an implementation of the equilibrium semantics for heterogeneous and nonmonotonic multi-context systems (MCS) [3], which feature contexts with heterogeneous and possibly nonmonotonic logics. Each context in an MCS comprises of two parts: a local knowledge base and a set of bridge rules that can access the beliefs of other contexts and add new information to the knowledge base. In this setting, contexts are loosely coupled, and may model distributed information linkage applications, thus it is natural to have a system that allows for the distributed evaluation of MCS. In an MCS M = (C_1..., C_n), each context C_i is characterized by a knowledge base kb_i and a set of bridge rules br_i. In our implementation, each kb_i is in DLV syntax as in [6]. The br_i are sets of nonmonotonic rules p_0 ← (c_1 : p_1)..., (c_j : p_j), not (c_(j+1) : p_(j+1))..., not (c_m : p_m). where the (c_k : p_k) are bridge atoms, the index c_k refers to a context C_(C_k) and p_k is a possible belief of C_(C_k), intuitively, the atom is true if p_k is in the belief set of context C_(C_k). If the body evaluates to true with respect to a belief state, which is a sequence S = (S_1..., S_n) of belief sets S_i of C_i, 1≤i≤n, then p_0 has to be added to kb_i. The semantics of M is then given in terms of stable belief sets (called equilibria). Partial Equilibria are equilibria in a sub-MCS of M induced by a single context C_k resp. a collection C_(k_1)..., C_(k_j), of contexts.
展开▼
