DFA Analysis Glossary

An overall assessment of code complexity based on multiple structural factors. This metric provides a quick indication of how difficult a module may be to understand, maintain, or modify.

The maximum level of nested control structures (IF/THEN/ELSE, loops, SELECT/CASE) found in the code. Higher nesting depths indicate more complex decision trees and can make code harder to follow and test.

The count of all conditional statements (IF/THEN/ELSE blocks) in a module. A high number of conditions increases the number of possible execution paths, making testing and validation more challenging.

Specific factors that contribute to the overall complexity assessment. These are identified during structural analysis and help pinpoint areas that may need attention.

Recurring structures that govern how code execution flows through a module. The analysis identifies these patterns to understand the logical structure and decision-making processes within the code.

How variables are categorized and organized within the code. This classification helps understand the purpose and scope of different data elements.

Types of value assignments found in the code. These patterns reveal how data is manipulated and transformed.

High-level design approaches identified in the code that reveal the overall architecture and design philosophy.

The degree of interdependence between a module and other parts of the system. Lower coupling generally indicates more maintainable and flexible code.

Links from a module to other components in the system. These references create the dependency graph and affect how changes propagate through the codebase.

An assessment of how complex the dependency graph is for a module. This considers not just the number of dependencies, but also their depth and interconnection.

Document Format definitions that specify how content is laid out and formatted within a document. DOCFORMATs contain the actual document structure, text placement, variable references, and formatting instructions.

Format Groups define the physical page structure including sheets, logical pages, and how DOCFORMATs are arranged. They act as containers that organize multiple formats into a cohesive document structure.

Reusable code modules that can be incorporated into other modules. INCLUDEs promote code reuse by centralizing common functionality, variable definitions, or initialization routines.

A clear, non-technical description of what the module accomplishes from a business perspective. This helps stakeholders understand the role of each module without needing to read the code.

The module's position and function within the overall document generation pipeline. This describes how the module interacts with other components and when it is invoked.

A description of how data moves through the module, from input to output. This includes data sources, transformations applied, and where results are stored or passed.

Specific business logic constraints and requirements implemented in the code. These rules govern how data is processed, validated, or formatted according to business requirements.

Specific locations where the module connects with other components, external systems, or data sources. Understanding integration points is crucial for impact analysis and testing.

A simplified, human-readable algorithmic representation of the module's logic. Pseudo-code abstracts away DFA syntax details while preserving the essential logic flow, making it easier for non-DFA developers to understand the code.

Problems or concerns identified during analysis that may affect code quality, maintainability, or reliability. Issues are categorized by priority level.

Specific suggestions for improving code quality, structure, or maintainability. These are actionable items that developers can implement to enhance the module.

Observations about code patterns that may impact execution speed, memory usage, or resource consumption. These help identify optimization opportunities.

Important information for developers who will maintain or modify the code in the future. These notes highlight non-obvious behaviors, dependencies, or historical context.

A composite score from 0 to 100 that evaluates the quality of LLM-generated analysis for each section. It combines four weighted dimensions: Depth (25%), Fidelity (30%), Specificity (20%), and Completeness (25%). Higher scores indicate more thorough, accurate, and detailed analysis outputs.

A letter grade derived from the quality score, providing a quick visual indicator of analysis quality.

Measures the depth of the analysis relative to the complexity of the source code. Evaluates the ratio of pseudo-code length to source code, description detail, and business logic coverage. A shallow analysis of complex code scores low, while a thorough analysis of simple code scores high.

Measures factual accuracy of the analysis based on post-validation confidence. Cross-references LLM outputs against the actual source code to verify dependencies, variables, and complexity claims. The highest-weighted dimension because accuracy is critical for documentation trustworthiness.

Detects the absence of generic or boilerplate descriptions. Scans analysis outputs for common generic patterns (e.g., "standard implementation", "typical approach") and penalizes their presence. Higher specificity means the analysis provides concrete, module-specific insights rather than generic observations.

Measures the percentage of analysis fields that are substantively filled. Checks 9 key fields: description, primary_function, business_logic, pseudo_code, data_flow, dependencies, potential_issues, recommendations, and integration_points. Empty or placeholder values reduce the score.

Automatic alerts triggered when specific quality thresholds are not met. These flags help quickly identify sections that may need re-analysis or manual review.

A cross-section coherence score (0-100%) that evaluates consistency across all analyzed sections. Checks for orphan dependencies (references to non-existent sections), missing back-references, and include consistency. A high coherence score indicates that cross-section relationships are accurately documented.