AcademicLibrary

A professional research tool for the local analysis and visualization of academic libraries. AcademicLibrary processes PDF documents to extract metadata and creates an interactive, force-directed network map based on bibliographic coupling.

Overview

The application utilizes the GROBID (Geneva Reference Extraction from Bibliographic Information Document) server to parse full-text PDFs. It identifies shared references between documents to calculate a similarity metric, visually clustering related research while maintaining a local, private database.

Core Features

1. Document Extraction

• Full-text parsing of academic PDFs using GROBID REST API.
• Metadata consolidation to verify titles, authors, and publication years against external records.
• Extraction of bibliography lists to enable similarity mapping.

2. Network Visualization

• Bibliographic Coupling: Nodes are connected based on shared citations in their respective bibliographies.
• Force-Directed Layout: Implementation of the ForceAtlas2 algorithm to distribute papers based on connection strength.
• Node Scaling: Geometric scaling of nodes according to their degree of connectivity within the local library.

3. Interactive UI

• Dual Sidebar Interface: A left-hand navigation pane for a full library list and a right-hand detail pane for metadata.
• Dynamic Detail View: Real-time updates of paper abstracts, full author lists, and years upon node hover.
• Stability Control: Automated physics stabilization that locks nodes in place once equilibrium is reached.

Technical Stack

• Processing Engine: GROBID (Dockerized).
• Network Analysis: NetworkX.
• Visualization Layer: Pyvis (D3.js based).
• Data Management: Pandas.

Implementation Details

Similarity Logic

The relationship between two papers ($i$ and $j$) is determined by the intersection of their reference sets: $$Strength = |Refs_i \cap Refs_j|$$ Connections are established for any non-zero intersection, pulling related papers into thematic clusters.

Metadata Processing

The tool implements a caching mechanism to avoid redundant processing. Extracted XML files are stored in a separate directory (Extracted_XML), ensuring the original PDF folder remains organized.

Setup Requirements

1. Docker: Ensure the GROBID server is running locally: docker run -t --rm -p 8070:8070 grobid/grobid:0.8.0
2. Directory Structure: Place all target PDFs in the designated /Papers directory.
3. Dependencies: Requires grobid-client-python, pandas, networkx, and pyvis.

Usage

Execute the main application script: python app.py

Upon completion, an interactive HTML file (connected_papers_sidebar.html) will be generated and automatically opened in the default system browser.