SEER adds spectral-entropy role encoding from Laplacian spectra and empirically calibrated time-weighted calling contexts to raise macro-F1 from 92.47% to 93.20% and accuracy from 92.52% to 93.98% on PyDesignNet for 23 GoF patterns.
https://developer.mozilla.org/en-US/docs/Web/API/Web_Workers_API/Structured_clone_algorithm (2025)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SE 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
SEER: Spectral Entropy Encoding of Roles for Context-Aware Attention-Based Design Pattern Detection
SEER adds spectral-entropy role encoding from Laplacian spectra and empirically calibrated time-weighted calling contexts to raise macro-F1 from 92.47% to 93.20% and accuracy from 92.52% to 93.98% on PyDesignNet for 23 GoF patterns.