Providing an Agent-Based Architecture for Semantic Mining From Large-Scale Data in Distributed Environments

Large-scale data may consist of big, distributed, scattered, heterogeneous, irrelevant, misleading, real, and unrealistic data or any combination of them. Therefore, analyzing, creating value and data productivity is always an important and open challenge. Therefore, the purpose of this study is to present a new coalition architecture for generating valuable information for decision making among the masses of data. The proposed architecture, abbreviated ASMLDE, aims to develop and improve data mining and semantic exploration, and to produce useful and high-quality rules consisting of four layers, seven components and six key elements. In the proposed architecture, conceptualization with 4v's process, insight into the volume and scale of data in the form of 3v's model and finally qualitative insight based on data thickness, are used for conceptualization and standardization of qualitative processes and more complex interpretations. This architecture, supported by ontology and agent mining, reduces large search spaces and increases the speed and quality of data mining operations due to the use of multi-agent systems. Automating exploration operations, reducing data complexity and business processes are also important achievements of the proposed architecture. To evaluate the proposed architecture, a large-scale dataset of natural disasters and earthquake ontology classes from the DBpedia knowledge base have been used. The evaluation results obtained by exploring the semantic rules of the mentioned dataset highlight the effectiveness and capabilities of the ASMLDE  architecture in enhancing the quality of the semantic rules explored to fit the user need and reducing the large data mining space over other similar frameworks and architectures.