README.md

metadata

title: ScholarFlow Research Intelligence Studio
emoji: 📚
colorFrom: blue
colorTo: green
sdk: static
app_file: index.html
pinned: false
license: mit
tags:
  - research
  - literature-review
  - evidence-synthesis
  - multilingual
  - reproducibility
  - static

ScholarFlow Research Intelligence Studio | Hugging Face Project #19

A privacy-conscious, local-first research evidence dashboard that turns an unstructured list of studies into an inspectable library, human-reviewed screening decisions, an evidence matrix, descriptive research-coverage diagnostics, and reproducible exports.

Owner / portfolio: Hugdrigo
Project: 19 of the Hugging Face portfolio series
Release: 1.0.0
Hosting: Hugging Face Static Space, free tier, no server dependency

Run

Open index.html in a current browser, or upload all files at the repository root of a Hugging Face Space configured as Static. No Python backend, paid inference, API keys, or Node packages are required.

Features

1. Overview with summary metrics and method/decision distributions.
2. Keyword-searchable and filterable study library.
3. Add sources manually, with methodological metadata and traceable origin fields.
4. Manual screening: include, maybe, exclude, and unscreened, with review notes.
5. Evidence matrix focused on included studies.
6. Descriptive distribution and threshold-based coverage insights.
7. JSON, CSV, and Markdown exports plus JSON import.
8. Browser-local persistence with explicit replacement and deletion confirmation.
9. Responsive layout for desktop and mobile, with no remote data collection.

Crucial integrity note

The included demonstration dataset is entirely synthetic. Records marked SYN-*, their authors, samples, findings, and decisions are illustrative and must not be cited as authentic research. User-entered records are not independently verified by the app. The tool does not query scholarly databases or execute generative-AI models. The term research intelligence refers to structured, rule-based decision support, not autonomous research or external LLM inference.

The completeness field counts populated metadata categories (source, methods, findings, limitations, participants), not scientific quality. Gap insights identify patterns in the user's included records only, not established gaps in the world's research literature.

Files

• index.html: Space entry point and semantic UI.
• styles.css: accessible responsive design.
• seed.js: demonstration data loaded without remote requests.
• sample-data.json: portable, clearly marked synthetic sample dataset.
• app.js: library, screening, visualization and import/export logic.
• METHODOLOGY.md: interpretation, screening process, reproducibility safeguards.
• LICENSE: MIT license.

Use with authentic literature

1. Back up your current state with Export full JSON.
2. In Export & methods, select Clear all records or import your own JSON.
3. Collect authentic sources from databases or publisher websites outside the app.
4. Add records and verify each bibliographic reference yourself.
5. Establish an a priori inclusion/exclusion protocol; record reasons in the screening desk.
6. Analyze only relevant, verifiable, included studies; export the resulting audit log.

JSON shape

{
  "records": [
    {
      "id": "REAL-001",
      "title": "Verified paper title",
      "authors": "Author Names",
      "year": 2025,
      "method": "Qualitative",
      "region": "Latin America",
      "language": "Spanish",
      "participants": 25,
      "topic": "academic writing",
      "findings": "Source-verified summary",
      "limitations": "Limitations reported by authors",
      "source": "doi:10.xxxx/example",
      "synthetic": false,
      "status": "unscreened",
      "notes": ""
    }
  ]
}

Use an actual, verified DOI rather than the placeholder above. Importing JSON replaces the existing dataset after confirmation. The app has no account sync: localStorage is device/browser-specific, so export frequently.

Next possible release

Optional integrations for Crossref/OpenAlex metadata, DOI verification, a documented review protocol, RAG retrieval, and rubric-based critical appraisal could be built as separately disclosed additions. These capabilities are not implemented in v1.0.

Attribution

Created as Project #19 for the public Hugdrigo Hugging Face professional portfolio. MIT licensed; see LICENSE.