Automated academic paper metadata & variable extraction with Gemini to Google Sheets

📚In the social and behavioral sciences (e.g., psychology, sociology, economics, management), researchers and students often need to normalize academic paper metadata and extract variables before any literature review or meta-analysis.

🧩This workflow automates the busywork. Using an LLM, it processes CSV/XLSX/XLS files (exported from WoS, Scopus, EndNote, Zotero, or your own spreadsheets) into normalized metadata and extracted variables, and writes a neat table to Google Sheets.

🔗 Example Google Sheet: click me

👥 Who is this for?

• 🎓 Undergraduate and graduate students or researchers in soft-science fields (psychology, sociology, economics, business)
• ⏱️ People who don’t have time to read full papers and need quick overviews
• 📊 Anyone who wants to automate academic paper metadata normalization and variable extraction to speed up a literature review

⚙️ How it works

1. 📤 Upload an academic paper file (CSV/XLSX/XLS) in chat.
2. 📑 The workflow creates a Google Sheets spreadsheet with two tabs: Checkpoint and FinalResult.
3. 🔎 A structured-output LLM normalizes core metadata (title, abstract, authors, publication date, source) from the uploaded file and writes it to Checkpoint; 📧 a Gmail notification is sent when finished.
4. 🧪 A second structured-output LLM uses the metadata above to extract variables (Independent Variable, Dependent Variable) and writes them to FinalResult; 📧 you’ll get a second Gmail notification when done.

🛠️ How to set up

🔑 Credentials

• Google Sheets OAuth2 (read/write)
• Gmail OAuth2 (send notifications)
• Google Gemini (or any LLM you prefer)

🚀 Quick start

1. Connect Google Sheets, Gmail, and Gemini (or your LLM) credentials.
2. Open File Upload Trigger → upload your CSV/XLSX/XLS file and type a name in chat (used as the Google Sheets spreadsheet title).
3. Watch your inbox for status emails and open the Google Sheets spreadsheet to review Checkpoint and FinalResult.

🎛 Customization

• 🗂️ Journal lists: Edit the Journal Rank Classifier code node to add/remove titles. The default list is for business/management journals—swap it for a list from your own field.
• 🔔 Notifications: Replace Gmail with Slack, Teams, or any channel you prefer.
• 🧠 LLM outputs: Need different metadata or extracted data? Edit the LLM’s system prompt and Structured Output Parser.

📝 Note

• 📝 Make sure your file includes abstracts. If the academic paper data you upload doesn’t contain an abstract, the extracted results will be far less useful.
• 🧩 CSV yields no items? Encoding mismatches can break the workflow. If this happens, convert the CSV to .xls or .xlsx and try again.

📩 Help

Contact: [email protected]