Goal Modeling¶
Goal modeling represents intended outcomes and their relations so stakeholders can examine refinements, dependencies, responsibilities, and system alternatives.
Core Idea¶
Goal modeling represents intended outcomes and their relations so people can examine possible means, dependencies, responsibilities and requirements. A KAOS goal graph and an i* actor model both make goals inspectable, but their link types have different meanings. Drawing a model neither proves a goal has been achieved nor mechanically produces a verified requirement.[ref-b1e2035ac87f][ref-ad4bb07f7931]
Scope of Application¶
The shared roles are an objective, an explicit model, interpretable intentional relations, actor context where relevant, and reasoning toward choices. Van Lamsweerde uses KAOS for BART train-control requirements; Yu uses i* to compare meeting-scheduling arrangements. AND/OR refinement belongs to the KAOS case, while softgoals and actor dependencies belong to the i* account; neither notation's full grammar is required of every goal model.[ref-b1e2035ac87f][ref-ad4bb07f7931]
Clarity¶
The model separates what outcome is wanted from which means are proposed and what a link asserts. In KAOS, a refinement relates broader and narrower goals under that method's conditions. In i*, dependencies and contributions express actor interests and qualitative alternatives. A positive contribution is a modeled judgment, not a measurement of operational success.[ref-b1e2035ac87f][ref-ad4bb07f7931]
Manages Complexity¶
A goal model keeps the route from a broad objective to narrower conditions, possible assignments and alternatives visible. BART's SafeTransport can be inspected alongside train-entry, spacing and speed-limit goals. Yu's meeting case exposes interests of initiator and participants before comparing a possible scheduler. The graph organizes questions for analysis while further validation remains necessary.[ref-b1e2035ac87f][ref-ad4bb07f7931]
Abstract Reasoning¶
To evaluate a proposed instance, find its goals, representation medium, link semantics, relevant actors and intended use in comparing means or requirements. A disconnected wish list has no relational model. An early model without actor allocation can still be a goal model, but it cannot yet justify responsibility claims. A proposed system alternative must be assessed separately from the model describing it.[ref-b1e2035ac87f][ref-ad4bb07f7931]
Knowledge Transfer¶
The practice of representing goals and meaningful links transfers from train safety to meeting organization. The concrete objectives and link rules must be rebuilt in each setting. The underlying target-to-medium mapping belongs to the Representation Prime, which goal modeling presupposes; the named practice remains a requirements-engineering specialization.[ref-b1e2035ac87f][ref-ad4bb07f7931]
Example¶
KAOS BART case. Van Lamsweerde's preliminary graph includes SafeTransport and conditions for avoiding train entry through a closed gate, maintaining train separation and respecting segment speed limits. Later refinement and allocation distinguish software requirements from environment expectations. The roles map as BART safety objective, KAOS graph, refinement links, later software/environment agents, and candidate requirements or expectations. The graph is not itself proof of train safety.[^ref-b1e2035ac87f]
i* meeting case. Yu's Figure 3 models an existing initiator-and-participant scheduling routine before a computer scheduler is considered. Figure 4 models a proposed computer-supported configuration with a scheduler actor, alternative means and softgoal contributions. The roles map as agreeable meeting objective, i* rationale/dependency medium, intentional links, initiator/participants/proposed scheduler, and comparison of alternatives. Neither figure reports deployment or measured success.[^ref-ad4bb07f7931]
Relationships to Other Abstractions¶
Current abstraction Goal Modeling Domain-specific
Parents (1) — more general patterns this builds on
-
Goal Modeling presupposes Representation Prime
Goal modeling requires a mapping of intended objectives and relations into an inspectable model.
Hierarchy path (1) — routes to 1 parentless root
- Goal Modeling → Representation → Abstraction
Neighborhood in Abstraction Space¶
Goal Modeling sits in a sparse region of the domain-specific corpus (95th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.
Family — Unclustered & Miscellaneous (2551 abstractions)
Nearest neighbors
- Event Correlation — 0.79
- Cradle-to-Cradle Design — 0.78
- Transmediation — 0.78
- Protocol Stack — 0.77
- Rational Choice Model — 0.77
Computed from structural-signature embeddings · 2026-10-08
Not to Be Confused With¶
A bare objective list, task flow without goals, or a Goal Structuring Notation assurance claim-and-evidence argument. These may share words or diagrams but not the two sources' intentional goal-modeling relations. The approved DAG relation is composition/presupposes to Representation: the practice requires a model, but is not itself only a representation artifact. Requirements Analysis is a nearby broader practice, not an additional established strict parent here.[ref-b1e2035ac87f][ref-ad4bb07f7931]
References¶
[^ref-b1e2035ac87f]: Axel van Lamsweerde, Goal-Oriented Requirements Engineering, A Guided Tour, invited mini-tutorial in Proceedings of the 5th IEEE International Symposium on Requirements Engineering (RE'01, 2001), pp. 249–263; original title-page punctuation is “Goal-Oriented Requirements Engineering: A Guided Tour.” See §§2–3 for goals/refinement and §6, Fig. 1 for the BART graph and responsibility analysis. The author-hosted PDF is the consulted full text. [^ref-ad4bb07f7931]: Eric S. K. Yu, Towards Modelling and Reasoning Support for Early-Phase Requirements Engineering, Proceedings of the Third IEEE International Symposium on Requirements Engineering (RE'97, 1997), doi:10.1109/ISRE.1997.566873, §§2.1–2.4, Figs. 1–4. Figure 3 models the pre-computer routine; Figure 4 models a proposed computer-supported configuration. The author-hosted PDF is the consulted full text.