Symbol & Notation Normalizer

Overview / 概述

Provides a standards-based conversion mapping and tooling to normalize symbols and notation extracted from technical documents into canonical forms. The core asset is a machine-readable mapping (raw/recognized form → standard symbol, with Unicode codepoint, standard reference, LaTeX, and ambiguity notes), plus a batch-normalization script and an offline HTML viewer.

提供一套基于标准的转换映射与工具，将技术文档（PDF、图片、Word 等）中识别出的符号与记法规范化为标准形式。核心资产是一份机器可读的映射表（识别形式 → 标准符号，含 Unicode 码位、标准出处、LaTeX 与歧义说明），外加批量规范化脚本与一个离线 HTML 查看器。

When to use

• Cleaning up symbols after OCR / document recognition (PDF, images, Word).
• Reviewing or authoring technical documents that require correct notation per ISO/IEC/IUPAP.
• Resolving ambiguous glyphs: U (voltage quantity) vs V (volt unit), Q (reactive power) vs Ω (ohm), ⌀ (diameter) vs Φ (phi), θ vs 0, μ vs u, j (electrical imaginary unit) vs i (math).
• Building or extending a notation-conversion pipeline.

Workflow

1. Identify the document domain (electrical / mechanical / math / physics / general-units) before mapping — the same glyph has different meanings per domain.
2. For lookup or display: open assets/symbol_display.html (offline, searchable, filterable viewer with the full mapping), or read assets/symbol_conversion_map.json directly.
3. For batch normalization of recognized text: run scripts/normalize.py (see its --help). It loads the JSON and replaces unambiguous misrecognitions (symbols, units, multi-character words) while leaving risky single-letter Latin substitutions for manual/contextual handling.
4. Apply the global ambiguity rules (R1–R6, see references/conversion_guide.md) for cases the script flags as ambig.
5. When generating output (LaTeX/MathML or final document), apply font rules: variables italic, units/functions/constants upright; mark superscripts and subscripts explicitly.

Resources

scripts/normalize.py

Batch normalization CLI. Modes: normalize a file/stdin/--text, --stats for dataset counts, --lookup with optional domain filter for querying the mapping. Safe by default: only auto-replaces non-single-letter misrecognitions; reports every change with its standard reference and any ambiguity (ambig field).

references/standards_summary.md

Categorized index of the standards behind each domain (ISO 80000, IEC 60027-x, IUPAP, ISO 128/129/1101, ASME Y14.5, SI), with scope and key symbols.

references/conversion_guide.md

Conversion principles, global disambiguation rules R1–R6, per-domain notes, and a suggested post-recognition pipeline.

assets/symbol_conversion_map.json

Canonical mapping. Each entry: id, domain, raw[], standard_symbol, codepoint, standard_ref, latex, note, ambig?. This is the single source of truth for both the script and the viewer.

assets/symbol_display.html

Self-contained offline viewer (data embedded) for web/WebView display; supports search, domain filter, and URL deep-links (?domain=electrical&q=Ω).

Notes

• Always prefer the ISO 80000 series (supersedes ISO 31). For Chinese projects, GB/T equivalents apply.
• The mapping is conservative; ambiguous items are flagged rather than force-replaced. Human review is recommended for flagged cases.