The World Wide Web contains a plethora of information that exists in structured, semi-structured, and unstructured formats. Furthermore, the same information can be textually described in many different ways. As a result, it creates challenges for automated systems in retrieving data from various sources and combining them into a single knowledgebase. To improve upon this process, we introduce SBIWS, a Semantics Based Intelligent Web Service architecture, which allows the user to create a customized Ontology based on metadata definitions which are used for gradually accumulating the extracted information into a domain-specific knowledgebase. SBIWS is a multi-agent system that uses software agents to configure user's subjects of interest, build up and refine a dictionary-like Metadata Repository according to users' requirements and preference, and grow this repository in an incremental learning process.