Get Help
Skip to Main Content

AI-Generated Tools for Academic Research

A guide that outlines the key features, benefits, drawbacks, and potential applications of AI-powered research tools.

Introduction to Connected Papers

More About Connected Papers

How it Works:

  • Uses a similarity metric combining: (a) Co-citations (two documents cited together by others) and (b) Bibliographic coupling (two works referencing a common third work)
  • Leverages the Semantic Scholar corpus:
    • A free, AI-powered research tool
    • Provides a large database of academic papers (around 200 million records)
    • Utilizes machine learning techniques

User Interaction:

  • Discover relevant papers by: a) Starting with one seed paper (can add more iteratively using multi-origin graphs) or b) Searching for keywords
  • Filter papers by publication year, keyword, PDF availability, or open access
  • Use list view to see percentage of similarity to origin paper
  • Use prior works to show seminal, most frequently cited papers
  • Use derivative works for recent papers

Visual Representation:

  • Generates a Force Directed Graph:
    • Each node represents a paper
    • Line length on map shows degree of connectedness
    • Size of the node represents number of citations
    • Darker color represents more recent publications
  • Clusters similar papers together (force-directed layout algorithm simulates attractive and repulsive forces between the nodes)

Notable Features:

Benefits:

  • Easy identification of key research areas
  • Highlights seminal works in the field
  • Intuitive visual interface for paper similarity exploration

Limitations:

  • Freemium model
    • Free version only allows for 2 graphs, then requires user account for up to 5 graphs per month
    • $6/month for academic subscription
  • Works best with English-language publications
  • Visual interface may challenge some users
  • Possible blind spots, such as:
    • underrepresented researchers and bias towards highly cited papers
    • niche topics or recent topics not cited much yet
    • articles behind a paywall or with abstract-only access
    • limited by the Semantic Scholar index, which has coverage gaps (particularly for monographs and textbooks) and occasional metadata completeness issues

Worksheet: Exploring Research Connections with Connected Papers