SemanticForge: Repository-Level Code Generation through Semantic Knowledge Graphs and Constraint Satisfaction

TL;DR

SemanticForge enhances code generation by integrating repository-wide semantic knowledge graphs with advanced algorithms, significantly reducing logical and schematic errors in large language model outputs.

Contribution

The paper introduces four novel algorithms for semantically-aware code generation, unifying semantics, improving query accuracy, enabling real-time constraint checking, and efficiently maintaining knowledge graphs.

Findings

Achieved 73% precision in graph query generation from natural language.

Developed a real-time constraint verification method during code generation.

Maintained semantic equivalence with $O(| riangle R| imes \log n)$ update time.

Abstract

Large language models (LLMs) have transformed software development by enabling automated code generation, yet they frequently suffer from systematic errors that limit practical deployment. We identify two critical failure modes: \textit{logical hallucination} (incorrect control/data-flow reasoning) and \textit{schematic hallucination} (type mismatches, signature violations, and architectural inconsistencies). These errors stem from the absence of explicit, queryable representations of repository-wide semantics. This paper presents \textbf{SemanticForge}, which introduces four fundamental algorithmic advances for semantically-aware code generation: (1) a novel automatic reconciliation algorithm for dual static-dynamic knowledge graphs, unifying compile-time and runtime program semantics; (2) a neural approach that learns to generate structured graph queries from natural language, achieving 73\% precision versus 51\% for traditional retrieval; (3) a novel beam search algorithm with integrated SMT solving, enabling real-time constraint verification during generation rather than post-hoc validation; and (4) an incremental maintenance algorithm that updates knowledge graphs in $O(|\Delta R| \cdot \log n)$ time while maintaining semantic equivalence.