From 39e3373c8782e9607b091e6aa632a20dcd950f61 Mon Sep 17 00:00:00 2001 From: masader-bot Date: Sun, 30 Aug 2026 00:39:59 +0300 Subject: [PATCH] Creating datasets/arams.json --- datasets/arams.json | 63 +++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 63 insertions(+) create mode 100644 datasets/arams.json diff --git a/datasets/arams.json b/datasets/arams.json new file mode 100644 index 00000000..27829a4d --- /dev/null +++ b/datasets/arams.json @@ -0,0 +1,63 @@ +{ + "Name": "AraMS", + "Volume": 28600.0, + "Unit": "documents", + "License": "CC BY-NC-SA 4.0", + "Link": "https://doi.org/10.5281/zenodo.22095333", + "HF_Link": "", + "Year": 2024, + "Source": [ + "books" + ], + "Form": "text", + "Domain": [ + "history" + ], + "Annotation_Style": [ + "LLM annotation", + "human annotation" + ], + "Description": "Historical Arabic manuscript dataset with line annotations.", + "Provider": [ + "Higher School of Computer Science (ESI-SBA)" + ], + "Derived_From": [ + "Muharaf" + ], + "Partial": false, + "Paper_Title": "RefLAM: A Reference-Grounded Line Annotation Pipeline for Historical Arabic Manuscripts", + "Paper_Link": "https://arxiv.org/pdf/2608.25140v1.pdf", + "Tokenized": false, + "Host": "GitHub", + "Access": "Free", + "Cost": "", + "Has_Splits": true, + "Tasks": [ + "other" + ], + "Venue_Title": "", + "Venue_Type": "preprint", + "Venue_Name": "", + "Authors": [ + "Mohamed Guechaoui", + "Mohamed Diaa Zellagui", + "Souleyman Chaib", + "Sahraoui Dhelim" + ], + "Affiliations": [ + "Higher School of Computer Science (ESI-SBA), Sidi Bel Abbes, Algeria" + ], + "Abstract": "The Arab-Islamic manuscript traditions span more than a millennium and cover medicine, astronomy, mathematics, philosophy, and jurisprudence. A large fraction of this corpus has never been rendered machine-readable. Photographic digitisation largely solved; the unsolved problem is transcription at scale. Handwritten-text recognition (HTR) models are data hungry, yet line-level labelled corpora for Arabic manuscripts remain scarce. Producing line-level OCR training data conventionally requires three expensive manual stages: page segmentation, transcription, and quality verification. For historical Arabic manuscripts, four well-documented difficulties compound this cost: high script variability (Naskh, Ruq\u2018ah, Thuluth, Maghrebi), degraded scan quality, non-linear two-zone reading order (main body plus margins), and diacritisation inconsistency. We measure the cumulative manual annotation rate directly: a trained annotator completes roughly 40 lines per hour, including line detection, transcription, and bounding-box drawing\u2014 a rate at which a single 600-page volume consumes the better part of a year. We address this bottleneck with RefLAM, which exploits two resources that, for many classical Arabic texts, already exist independently of manuscript images: (1) a vision-capable MLLM that produces structured, layout-tagged OCR from a single zero-shot call; and (2) a clean ground-truth transcription\u2014 typically a fully diacritised scholarly edition. Neither resource alone solves the problem: MLLM output is fluent but unreliable; the reference text is reliable but not localised to any page image. RefLAM fuses them via diacritic-agnostic fuzzy alignment, attaching a confidence score to every line and triaging human review accordingly. RefLAM is best read as an instance of weak (distant) supervision: an noisy labelling source (MLLM OCR) is reconciled against a reliable but unlocalised one (the reference transcription) until the labels are both accurate and grounded in the page; section 2 places the pipeline in that literature. The alignment stage has one property we lean on throughout: a maximal similarity score is not a heuristic signal but a provable guarantee of character-for-character normalised-string identity (proposition 1). We noticed the pattern before we proved it \u2014 during page-level review of the earliest books, no perfect-scoring line was ever found to be wrong\u2014and the proof then explained the observation. This confidence-100 rule is what makes the fast review path safe; it is an enabling property of the alignment stage, not a contribution that competes with the pipeline itself. Although we instantiate and validate RefLAM on Arabic manuscripts, the pipeline places no Arabic-specific requirement beyond the normalisation operator of definition 1; in principle it should transfer to any historical script for which clean digital transcriptions exist.", + "Dialect_Subsets": [ + { + "Name": "AraMS", + "Volume": 28600.0, + "Unit": "documents", + "Dialect": "mixed" + } + ], + "Dialect": "mixed", + "Language": "ar", + "Script": "Arab", + "Added_By": "qwen/qwen3.6-35b-a3b" +} \ No newline at end of file