Reconstructing Fragmentary Inscriptions with M-RADAR: Cross-Inscription Computational Analysis of the Singapore Stone and the Karimun Inscription

The decipherment of fragmentary inscriptions remains a major challenge in
epigraphy, historical linguistics, and digital paleography, particularly when
the surviving textual evidence is severely degraded or incomplete. The
Singapore Stone, one of Southeast Asia’s most enigmatic inscriptions, has
resisted definitive interpretation for nearly two centuries due to its
fragmentary condition and the absence of a systematic computational
framework for comparative analysis. This study uses M-RADAR (Maritime
Reconstruction via Automated Digital Analysis and Restoration), a
computational epigraphy framework that facilitates cross-inscription
morphological comparison and predictive reconstruction. Using the betterpreserved Karimun Inscription as a morphological reference corpus, the
framework combines high-resolution noise reduction, vector-based
graphemic analysis, nearest-neighbor classification, and probabilistic
syntactic modeling. The study reveals a 78.4% morphological
correspondence between the two inscriptions and achieves an 89%
predictive reconstruction accuracy for damaged textual segments. In
addition, it identifies previously unrecognized diacritic markers and
supports a Sanskrit-Kawi hybrid linguistic structure as the most probable
interpretation of the inscription. These findings provide new empirical
evidence for a shared paleographic tradition within the maritime Malay
Archipelago and demonstrate the effectiveness of cross-inscription computational reconstruction. More broadly, this study contributes a reproducible methodological framework for the analysis, restoration, and
digital preservation of fragmentary epigraphic materials in Southeast Asia.