🕷️️ Job Radar • SCRAPING

upwork.com 🟠 2026-05-10

🔹 Technical Manifest
👤 Client: 🇹🇷 Turkey Member since 2025-08-04
💰 Price: ****
🚩 Problem: Extract structured spare parts data from 100 large, multi-page PDF manuals with varying layouts.
📦 Existing: Not specified

Specifications:

[Target] Extract Part Numbers, Names, Descriptions, and Remarks from 100 PDFs each containing up to 500 pages.
[Method] Use OCR (Optical Character Recognition) for text extraction followed by NLP (Natural Language Processing) for structured data parsing.
[UI/UX] Not applicable
[Stack] Python with libraries like PyMuPDF, Tesseract-OCR, and NLTK. Database management using SQLite or PostgreSQL.
[Security] Ensure data privacy and security during processing; use encryption where necessary.
[Format] Output in CSV or Excel format with columns: part_number, part_name, source_pdf, remark.

Workflow:

1. Preprocess PDFs by converting them to a searchable format using PyMuPDF for better OCR accuracy.
2. Implement OCR using Tesseract-OCR to extract text from each page of the PDFs.
3. Develop NLP models to parse and structure extracted data into part numbers, names, descriptions, and remarks.
4. Handle table variations by training custom NLP models for different template structures.
5. Validate and clean the structured data to ensure high accuracy and no skipped extractable data.
6. Export processed data in CSV or Excel format with specified columns.