Research Article Text Mining for Meta-Analysis
Budget: $30 – $250 USD
Dear Estemeed Experts,
I am releasing a Request for Proposal (RFP) for the following work.
Title: Automated Text Mining and Meta-Analysis Dataset Creation from 85 Research Articles
Project Overview
I am conducting a meta-analysis of carefully selected reserach papers.
The goal is to automatically extract structured quantitative data from 85 peer-reviewed academic papers (PDFs) and create a meta-analysis-ready dataset in CSV/Excel format.
Scope of Work
Phase 1 – Text Mining and Data Extraction
The contractor will:
• Parse 85 PDF research articles (provided) using text-mining or semi-automated extraction.
• Identify and extract key fields from each paper’s abstract, methods, results, and tables.
• Populate a standardized meta-analysis table with the following variables:
study_id, effect_id, authors, year, title, doi,
region, country_sample, asset_class, method,
outcome_type, outcome_scale,
yi, sei, vi, direction_original,
where_found, extraction_mode, extraction_confidence, rob_overall
Expected Output:
meta_85_master.csv – fully populated meta-analysis dataset, validated and traceable (where_found = table/page reference).
As an additional deliverable, the contractor may run the full meta-analysis and produce all statistical outputs and figures using the provided R/Python scripts.
Tasks:
• Execute REML random-effects models (metafor / statsmodels).
• Perform multilevel and RVE (robust variance) meta-regressions.
• Conduct heterogeneity, publication bias, PET-PEESE, Egger tests.
• Generate publication-ready figures (forest, funnel, PET-PEESE, influence plots).
• Export final results to:
o tables/ (main, moderator, bias, subgroup)
o figures/ (forest, funnel, contour, PET-PEESE)
o report/ (HTML + PDF)
Item Description
1 meta_85_master.csv – standardized dataset
2 validation_report.csv – duplicate/missing/format check
3 (Optional) Complete analysis bundle (artifacts/) with all tables, figures, and reports
4 Short script (validate.py) confirming dataset consistency (vi=sei², no duplicates)
Provided Materials (for selected applicants)
Qualified applicants will receive:
• 85 PDF research papers
• Excel file with initial metadata (titles, DOIs, and base fields)
• Full bilingual “Meta-Analysis Data Extraction Protocol” and Execution Specification (R + Python)
Requirements
• Experience with text mining, NLP, or data extraction from academic PDFs
• Proficiency in Python (pandas, regex, pdfminer, Camelot, PyMuPDF) or R
• Understanding of meta-analysis methods (REML, PET-PEESE, RVE) preferred
• Ability to ensure reproducible outputs and proper validation
• Optional: capacity to run and deliver all analytical results (R/Python)
NDA will also be a requirement to sign!
Submission
Please include:
• A brief description of your experience with academic data extraction or meta-analysis.
• Whether you can also deliver the Phase 2 (full meta-analysis execution).
• Tools or frameworks you plan to use (e.g., Python, R, NLP models, OCR/LLM).
• Basic offer for the text mining and if possible, an optional offer for the statistical output.
Note: Only shortlisted applicants will receive the complete package (85 PDFs + Excel + specifications).
All work must ensure data traceability and reproducibility.
I am releasing a Request for Proposal (RFP) for the following work.
Title: Automated Text Mining and Meta-Analysis Dataset Creation from 85 Research Articles
Project Overview
I am conducting a meta-analysis of carefully selected reserach papers.
The goal is to automatically extract structured quantitative data from 85 peer-reviewed academic papers (PDFs) and create a meta-analysis-ready dataset in CSV/Excel format.
Scope of Work
Phase 1 – Text Mining and Data Extraction
The contractor will:
• Parse 85 PDF research articles (provided) using text-mining or semi-automated extraction.
• Identify and extract key fields from each paper’s abstract, methods, results, and tables.
• Populate a standardized meta-analysis table with the following variables:
study_id, effect_id, authors, year, title, doi,
region, country_sample, asset_class, method,
outcome_type, outcome_scale,
yi, sei, vi, direction_original,
where_found, extraction_mode, extraction_confidence, rob_overall
Expected Output:
meta_85_master.csv – fully populated meta-analysis dataset, validated and traceable (where_found = table/page reference).
As an additional deliverable, the contractor may run the full meta-analysis and produce all statistical outputs and figures using the provided R/Python scripts.
Tasks:
• Execute REML random-effects models (metafor / statsmodels).
• Perform multilevel and RVE (robust variance) meta-regressions.
• Conduct heterogeneity, publication bias, PET-PEESE, Egger tests.
• Generate publication-ready figures (forest, funnel, PET-PEESE, influence plots).
• Export final results to:
o tables/ (main, moderator, bias, subgroup)
o figures/ (forest, funnel, contour, PET-PEESE)
o report/ (HTML + PDF)
Item Description
1 meta_85_master.csv – standardized dataset
2 validation_report.csv – duplicate/missing/format check
3 (Optional) Complete analysis bundle (artifacts/) with all tables, figures, and reports
4 Short script (validate.py) confirming dataset consistency (vi=sei², no duplicates)
Provided Materials (for selected applicants)
Qualified applicants will receive:
• 85 PDF research papers
• Excel file with initial metadata (titles, DOIs, and base fields)
• Full bilingual “Meta-Analysis Data Extraction Protocol” and Execution Specification (R + Python)
Requirements
• Experience with text mining, NLP, or data extraction from academic PDFs
• Proficiency in Python (pandas, regex, pdfminer, Camelot, PyMuPDF) or R
• Understanding of meta-analysis methods (REML, PET-PEESE, RVE) preferred
• Ability to ensure reproducible outputs and proper validation
• Optional: capacity to run and deliver all analytical results (R/Python)
NDA will also be a requirement to sign!
Submission
Please include:
• A brief description of your experience with academic data extraction or meta-analysis.
• Whether you can also deliver the Phase 2 (full meta-analysis execution).
• Tools or frameworks you plan to use (e.g., Python, R, NLP models, OCR/LLM).
• Basic offer for the text mining and if possible, an optional offer for the statistical output.
Note: Only shortlisted applicants will receive the complete package (85 PDFs + Excel + specifications).
All work must ensure data traceability and reproducibility.