Support for semantic content is becoming more common in Web-accessible information systems. We see this support emerging with the use of ontologies and machine-readable, annotated documents. The practice of domain modeling coupled with the extraction of domain-specific, contextually relevant metadata also supports the use of semantics. These advancements enable knowledge discovery approaches that define complex relationships between data that is autonomously collected and managed. The InfoQuilt (One of the incarnations of the InfoQuilt system, as applied to the geographic information as part of the NSF Digital Library II initiative is the ADEPT-UGA system [Ade]. This research was funded in part by National Science Foundation grant IIS-9817432.) system supports one such knowledge discovery approach. This paper presents (parts of) the InfoQuilt system with the focus on its use for modeling and utilizing complex semantic inter-domain relationships to enable human-assisted knowledge discovery over Web-accessible heterogeneous data. This includes the specification and execution of Information Scale (IScapes), a semantically rich information request and correlation mechanism.