Figure 1: The main interface includes a paper, the papers it references (left), the papers that cite it (right), abstract, top words and links to full text version. The user navigates by simply clicking on papers, or using search (top).

Figure 2: Additional exploration tools. Here, a force-directed interactive layout. Each edge joins researchers who co-authored at least 2 papers. See it in action! (Chrome recommended)
Who is this for?
Researchers who want to explore an area of research and make sure they're not missing anything.
What is this?
This is a free, open source, BSD-licensed tool that allows you to build a library of papers that are relevant to your research and to easily explore the related work. Additional tools include search, a listing of often-referenced papers you may have missed, and a graphical visualization of the main researchers in your area and their co-author networks. Structured data is scraped using Microsoft Academic Search API.
Interesting, may I see a demo?
Sure, have a look here (Chrome is fastest). This demo was seeded with a fixed library of some papers from Deep Learning literature. Here's another demo library seeded with some 3D reconstruction work.
How can I get this and start building my own library?
Beta version is available here. For now, the server and data are hosted locally and are composed of a few Python scripts and pickles, but eventually this could all be pushed into cloud.
I have awesome ideas for more features. Can I help?
Certainly, we welcome contributions on Github.
You may also be interested in our Google Group





About Us

Andrej Karpathy is a Computer Science PhD student at Stanford who is upset about how difficult it is to explore academic literature. Find me on Twitter or G+.