Code clone explanation paper accepted at APSEC 2025

Our paper on explaining code clone detection with large language models was accepted in the Early Research Achievements (ERA) track of APSEC 2025, the 32nd Asia-Pacific Software Engineering Conference, held from 2 to 5 December 2025 in Macao. Code clone detection models such as CodeBERT and GraphCodeBERT work as black boxes, and the paper proposes using a large language model to explain their decisions after the fact. Its Knowledge-based Local Neighbourhood (KLN) approach selects meaningful examples around a clone pair and lets the model write an explanation that people can read. The work grew from Chayanee Junplong's master's thematic paper, and Chayanee presented it at the conference in December.

The paper was accepted as “Towards Human-interpretable Explanation in Code Clone Detection using LLM-based Post Hoc Explainer” and appears in the proceedings under the title below.