Introduction
Land-use planning involves partitioning geographic areas into zones with specific permissions, such as "no fishing" or "shipping only." This process is inherently complex, requiring planners to balance competing political, environmental, and infrastructural demands. Traditional planning software often relies on "black box" optimization methods that maximize a "goodness" score but fail to explain why specific decisions were made or how to resolve underlying policy contradictions. This paper introduces SPECS, a deductive planning assistant designed to help human planners develop consistent, legally sound zoning maps through logic-based rule refinement.
The SPECS System
SPECS (Spatial Planning Expert Constraint System) shifts the focus from automated map generation to interactive rule development. At its core, the system utilizes a logic-based rule engine that employs backward chaining to determine the appropriate zone for any given region. Unlike traditional systems, SPECS treats zoning as a functional mapping problem: the goal is for every region to be assigned exactly one zone based on the established rules.
The system is integrated with a spatial database, allowing it to evaluate geometric predicates—such as a region's proximity to a coastline or shipping lane—as part of its logical antecedents. Furthermore, SPECS extends standard logic programming to handle complex real-world scenarios, including disjunctive consequences (where a region might belong to one of several possible zones) and constructive negation (defining what a region cannot be). It also addresses "soft" rules through an "As Far As Possible" (AFAP) logic, treating them as retractable assumptions to avoid hard conflicts.
Visualizing Policy Inconsistency
One of the primary advantages of the deductive approach is the ability to visualize where policies fail. By treating the zoning relation as a function, SPECS can identify regions that are either "under-determined" (no rules apply) or "conflicting" (multiple rules apply).
As shown in Figure 1, the system uses color-coding to highlight these problem areas. A region appearing in a "conflict color" immediately alerts the planner that two or more policies are competing for control over that specific coordinate. This allows for an iterative planning cycle where the human user modifies the rules or the data to resolve these discrepancies, rather than simply accepting an optimized but potentially contradictory output.
Transparency and the Explanation Sub-system
Because SPECS is based on formal deduction, every zoning assignment is backed by a clear logical path. This addresses the "black box" problem of previous optimization-based systems like LUPLAN.
Figure 2 illustrates the system's explanation sub-system. When a planner queries a specific region, SPECS provides a "derivation tree" that lists the specific rules and data tuples used to reach a zoning decision. This transparency is crucial for legal and administrative accountability, as it provides a clear justification for every boundary on the map. It also helps planners understand the "downstream" consequences of their policy changes in real-time.
Conclusion
The SPECS system demonstrates that a logic-based, deductive approach is computationally feasible for large-scale spatial planning, as evidenced by its application to the Cairns Section of the Great Barrier Reef. By framing planning as a process of rule refinement rather than simple optimization, SPECS empowers human planners to create more consistent, transparent, and defensible zoning plans. While challenges remain in formalizing aesthetic judgments and automating conflict resolution, the system provides a robust framework for managing the complexities of environmental and land-use policy.
Summary updated