The field of coding theory and software mining is experiencing significant advancements, driven by the integration of large language models (LLMs) and innovative coding techniques. Researchers are exploring new methods to incorporate code structural knowledge into LLMs, enabling improved code translation and error correction. One notable direction is the use of in-context learning to post-incorporate code structural knowledge into pre-trained LLMs, which has shown promising results in improving performance without requiring extensive retraining. Another area of focus is the development of error-correcting codes, including deletion-correcting codes and constrained codes for DNA data storage. These codes are being designed to correct errors and ensure reliable data retrieval, with techniques such as LLM-guided search and syndrome-like decoding being employed. Noteworthy papers include:

• Post-Incorporating Code Structural Knowledge into LLMs via In-Context Learning for Code Translation, which proposes a novel approach to integrating code structural knowledge into LLMs.
• LLM-Guided Search for Deletion-Correcting Codes, which leverages LLMs to construct large deletion-correcting codes and achieves state-of-the-art results.
• LOCO Codes Can Correct as Well: Error-Correction Constrained Coding for DNA Data Storage, which introduces a new class of constrained codes with error correction capabilities for DNA data storage.