面向中文标书（MD）的混合抽取：规则/NER 优先，仅将低置信或冲突片段交给 LLM——可审计、低成本、高效率。Hybrid extraction for Chinese tender docs (MD): rules/NER first, LLM only for low-confidence/conflicts—auditable, lower cost, higher efficiency.
nlp json-schema regex extraction text-analysis information-extraction openai procurement bidding chinese-text-segmentation ner markdown-converter deduplication document-processing structured-output pydantic llm ollama tender-extract
-
Updated
Aug 21, 2026 - Python