Function association is a key task for binary reverse engineers, but existing search tools that perform it at scale do not fully exploit AI-enabled search, according to a paper posted on arXiv.
The paper, 'Identifier-Free Code Embedding Models for Scalable Search,' proposes embedding models that work without relying on identifiers. This approach is designed to address the gap between current search tools and the full capabilities of AI-enabled search.
The abstract notes that prior work has explored this area but does not provide experimental details. The paper is available on arXiv under the identifier 2605.05251.