Methodology Practice and Model Quality Assessment
Domain 4 tests methodology-related practice: how a modeling process is organized and how model quality is assessed. Expect questions on OOSEM/Harmony-style flows, viewpoint-driven reviews, and measurable quality criteria rather than tool menus.
Process and model organization
A methodology defines activities, artifacts, viewpoints, and quality gates. OOSEM emphasizes stakeholder needs, system requirements, architecture, and verification, with SysML models as the integrating artifact. Harmony-SE follows requirements analysis, system functional analysis, and architecture design with iterative verification. Whatever the label, the exam looks for explicit entry and exit criteria, allocated responsibilities, and traceability from needs to verification.
Quality assessment
Assess syntax, semantics, and pragmatics. Syntax: profile conformance, well-formed diagrams, valid value types. Semantics: ports and flows type-compatible; activities and state machines consistent with structure; constraints dimensionally consistent. Pragmatics: completeness, consistency, traceability coverage, readability, and fitness for the stakeholder's question. Metrics include requirements without «satisfy», blocks without allocation, unconnected ports, orphan elements, naming-convention violations, and trace gaps. Automated OCL and validation rules plus peer review and simulation are stronger than diagram counts.
Worked example
Assess a model with 50 requirements. An OCL query returns 12 «requirement» elements with no incoming «satisfy» from a block; 3 are intentional because they are derived to another model, leaving 9 gaps. Add the missing allocations, re-run the query, and review trace coverage against the verification plan. Reasoning: the metric localizes gaps; human triage separates intentional from missing; closure requires evidence, not merely a lower count.
Traps
Equating diagram count with model quality; relying only on tool syntax validation; treating every trace as semantically valid; omitting library and profile versions from a quality baseline; using a methodology name without its exit criteria; ignoring stakeholder viewpoint when judging completeness.