SCLA: Automated Smart Contract Summarization via LLMs and Control Flow Prompt
This program is tentative and subject to change.
Smart contract code summarization facilitates maintenance and vulnerability detection. Although large language models (LLMs) are widely adopted, they underperform fine-tuned models such as CodeT5+ and CodeBERT. Existing data-flow-based methods often miss hierarchical and control structures, leading to information loss and lower summarization quality. We propose \textbf{SCLA}, a multimodal LLM approach that enhances code understanding via semantically enriched prompts. SCLA integrates Function Call Graphs (FCGs) and control-flow semantic facts extracted from the Abstract Syntax Tree (AST) to preserve structural and contextual dependencies, including both explicit and implicit behaviors. Evaluated on 40,000 real-world smart contracts, SCLA achieves substantial gains over state-of-the-art baselines: 26.7% (BLEU-4), 23.2% (METEOR), 16.7% (ROUGE-L), and 14.7% (BLEURT), demonstrating that control flow–enhanced prompting effectively bridges the gap between general-purpose LLMs and domain-specific code understanding.
This program is tentative and subject to change.
Thu 16 AprDisplayed time zone: Brasilia, Distrito Federal, Brazil change
14:00 - 15:30 | AI for Software Engineering 15New Ideas and Emerging Results (NIER) / Research Track at Europa II Chair(s): Christian Bird Microsoft Research | ||
14:00 15mTalk | SCLA: Automated Smart Contract Summarization via LLMs and Control Flow Prompt New Ideas and Emerging Results (NIER) Xiaoqi Li Hainan University, Yingjie Mao Hainan University, Zexin Lu Hong Kong Polytechnic University, Wenkai Li Hainan University, Zongwei Li Hainan University | ||
14:15 15mTalk | Leveraging Design-Aware Context in Large Language Models for Code Comment Generation New Ideas and Emerging Results (NIER) Aritra Mitra Indian Institute of Technology Kharagpur, Srijoni Majumdar University of Leeds, Anamitra Mukhopadhyay Indian Institute of Technology Kharagpur, Partha Pratim Das Ashoka University, Paul Clough University of Sheffield, Partha Pratim Chakrabarti Indian Institute of Technology, Kharagpur | ||
14:30 15mTalk | From Execution to Embedding: Enriching Code Representations with Data Difference Signals for Comment Generation New Ideas and Emerging Results (NIER) Giacomo Fantino Politecnico di Torino, Italy, Antonio Vetrò Politecnico di Torino, Marco Torchiano Politecnico di Torino, Federica Cappelluti Politecnico di Torino, Italy | ||
14:45 15mTalk | Towards Bridging Language Gaps in OSS with LLM-Driven Documentation Translation New Ideas and Emerging Results (NIER) Elijah Kayode Adejumo George Mason University, Mariam Guizani Queen's University, Canada, Fatemeh Vares George Mason University, Brittany Johnson George Mason University | ||
15:00 15mTalk | Automating API Documentation from Crowdsourced Knowledge Research Track Bonan Kou Purdue University, Zijie Zhou University of Illinois Urbana-Champaign, Muhao Chen University of Southern California, Tianyi Zhang Purdue University | ||
15:15 15mTalk | UniCoR: Modality Collaboration for Robust Cross-Language Hybrid Code RetrievalDistinguished Paper Award Research Track Yang Yang Central South University, China, Li Kuang Centrel South University, Jiakun Liu Harbin Institute of Technology, Zhongxin Liu Zhejiang University, Yingjie Xia Hangzhou Dianzi University, David Lo Singapore Management University | ||