Briefed a batch JPG OCR tool as a real queue/ZIP sibling of the single-file page, not a multiple-attribute doorway.

Date: 2026-09-08 08:51 Summary: Briefed a batch JPG OCR tool as a real queue/ZIP sibling of the single-file page, not a multiple-attribute doorway. Visibility: people

[question] 立项：batch convert jpg to text with ocr，需要额外的info Gain增益，区别于convert-a-jpg-to-text-with-ocr

[try to solve]

Context

The site already ships convert-a-jpg-to-text-with-ocr (OCR N1): one printed JPG/PNG/WebP, Convert, copy or download a single TXT, language chips, local Tesseract LSTM loaded after click. Its brief and FAQ treated batch ocr / bulk ocr (Keyword Planner 10–100) as a later page and kept multiple off so the single-file URL would not fake a bulk job.

The request was to open that deferred page, with a hard constraint: extra Information Gain versus N1. Google scaled-content / doorway rules forbid a second URL that only changes the H1. Planner analysis (docs/seo/keywords/ocr-scan/2026-09-07-ocr-scan-keyword-planner.md) already said batch OCR is a different job, not an absorb into N1.

Process

1. Read N1 02-tool-info.md, the OCR/Scan Planner, the 09-08 Text Converter Planner, and an existing batch converter brief (batch-convert-web-pages-to-jpg) for queue/ZIP/skip patterns.
2. Confirmed Planner evidence: batch ocr, bulk ocr (10–100); bulk ocr scanning solutions is a service query (FAQ only); pdf to ocr multiple document is scanned-PDF (N2), not this page. Head terms jpg to text / image to text stay on N1.
3. Wrote work-tasks/batch-convert-jpg-to-text-with-ocr/ (00–03, notes.md): slug/H1 = task sentence Batch convert JPG to text with OCR; coverage table + Ads/Planner table + intent review; ten locale briefs with local batch phrasing (not N1 keyword copies).
4. Ran npm run coverage:gate -- --slug=batch-convert-jpg-to-text-with-ocr --phase=0b (OK). Marked 02 ready, 03 briefs-ready. No catalog/Page/i18n in this session.
5. Updated the Planner line, theme README, keyword pool, and keyword-to-tool tracker so the old “defer batch” note does not contradict the named brief.

Root cause / analysis

N1 and a batch page share the same OCR engine. That is not enough to split URLs. The split is valid only if the primary control and artifact change:

If implementation only sets multiple on N1’s dropzone, the page is a doorway and should not ship. Single-file jpg to text must not become this page’s H1; FAQ should send one-photo users back to N1.

Solution

• Slug / H1: batch-convert-jpg-to-text-with-ocr / Batch convert JPG to text with OCR (Chinese direction: 用 OCR 批量把 JPG 转成文字).
• Default product: up to 10 images; Convert all; ZIP contains {stem}.txt plus combined.txt; Copy combined text; language chips apply to the whole batch; load-sample exception (do not auto-run WASM on landing).
• Related (when built): convert-a-jpg-to-text-with-ocr, extract-text-from-pdf. N1 FAQ “batch?” must later link here (CROSS_TOOL_UPDATE=1).
• Not in scope: PDF input, Word/Excel dropdowns, invoice fields, handwriting product, scanning-bureau services, parallel OCR workers.

Implementers should start from 02-tool-info.md and notes.md, then follow tool-creation (opts page, vendor Tesseract + fflate, verify:tool). This session stops at brief + 0b.

Notes / boundaries

• No live SERP snapshot; do not claim long_gap or weekly attack KPI.
• Engine reuse is N1 Tesseract LSTM already in /vendor/tesseract/; do not add Paddle VL for “more IG”.
• Paddle.js was not shipped on N1 (size); this brief does not reopen that choice.
• Work-todos item “batch scan” is a different object (often scanned PDF). This slug is photos → TXT ZIP only.

[actions]

• Added work-tasks/batch-convert-jpg-to-text-with-ocr/ (00–03, notes.md); 02=ready, 03=briefs-ready
• npm run coverage:gate -- --slug=batch-convert-jpg-to-text-with-ocr --phase=0b
• Updated docs/seo/keywords/ocr-scan/2026-09-07-ocr-scan-keyword-planner.md, docs/seo/keywords/ocr-scan/README.md, docs/seo/keyword-daily-pool.tsv, docs/seo/keyword-to-tool-tracker.md