1 January 2025 in IEEE Transactions on Instrumentation and Measurement
31 March 2026 in ACM Transactions on Software Engineering and Methodology
In the rapidly evolving landscape of open-source software, code cleanliness has become a critical metric for assessing project quality and maintainability. However, understanding its evolution is essential for identifying long-term sustainability and best practices. This study investigates trends in code cleanliness across representative projects from three major open-source communities: Apache, Google, and Spring. These were selected based on popularity and active commit frequencies. We focus on indicators such as cyclomatic complexity, lines of code per function and file, line length, and naming practices, drawn from project-specific coding guidelines and relevant for evaluating AI-generated code. Using abstract syntax trees and thresholds from best practice to parse positive and negative modifications and integrating differential information from commit logs, we establish precise mappings between functions and code lines before and after modifications. Our analysis shows complexity and lines of code growth as projects mature, largely due to feature growth and faulttolerance logic in distributed systems. Continuous refactoring by key contributors can curb this rise overall. Projects also differ in standards and statement use: conditional and exception-handling statements correlate with higher complexity, while variable declarations and expressions track method size and line length, suggesting disciplined statement choices support cleaner code in practice.
30 June 2026 in Proceedings of the ACM on Software Engineering
With the increasing complexity and scale of integrated circuit (IC) designs, automated circuit design methods are essential for Verilog implementation. Although Large Language Models (LLMs) perform well in general-purpose coding such as C++ and Python, their performance in Verilog is constrained by domain-specific semantic and structural rules as well as the scarcity of high-quality training datasets. To address these issues, this work proposes a semantically enhanced Verilog code parsing and automatic repair framework based on Abstract Syntax Tree (AST). First, an advanced register-transfer level (RTL) analysis framework was developed to achieve semantic enhancement through static analysis–driven functional role inference and attribute annotation. Second, enhanced AST information is used to construct structured prompts and generate semantically rich module descriptions via an Application Programming Interface (API) for dataset construction. Finally, an AST-guided automatic Verilog repair framework was designed, which leverages enhanced AST analysis for precise defect localization and intelligent repair through compiler feedback loops. Experimental results indicate that the proposed method successfully repaired 15.24% of defective Verilog code, resulting in a high-quality RTL benchmark dataset containing 318,021 samples. Models fine-tuned on this dataset demonstrate significant performance improvements across three benchmarks, achieving average improvements of 13.76% and 16.95% on Eval-Machine pass@1 and pass@5, 8.97% and 14.56% on Eval-Human pass@1 and pass@5, and 15.70% and 12.20% on RTLLM V1.1 Syntax-VCS and Func.