Agent-Based Parallelization of a Multi-Dimensional Semantic Database Model

Responses to database queries that may be even identical should vary if they are given under a different user context. For instance, queries for wild animals in the context of the ocean versus mountains should be different. Announced in 1993 [1], Mathematical Model of Meaning (MMM) provides users with capabilities to extract data items tightly coupled under different semantic spaces. Such a space is created dynamically with user-defined impression words to compute semantic equivalence and similarity between data items. MMM computes semantic correlations between the key and other data items to achieve dynamic data querying. However, a semantic space creation and a data correlative calculation are computationally demanding. We consider MMM as a practical database application of multi-agent technologies, construct a space over a cluster system, and have multi-agents explore for a given target and its surrounding data items. We use the Multi-Agent Spatial Simulation (MASS) library to implement an agent-based semantic database system and to measure its parallel execution. Compared to a sequential MMM implementation, MASS-based parallelization yielded a 22-time speedup when creating a space, mainly achieved with matrix multiplication. MASS also reduced the time required for distance sorting of multidimensional vectors by 23.7%.