NNexus: An Automatic Linker for Collaborative Web-Based Corpora

In this paper, we introduce Noosphere Networked Entry eXtension and Unification System (NNexus), a generalization of the automatic linking engine of Noosphere (at PlanetMath.org) and the first system that automates the process of linking disparate "encyclopediardquo entries into a fully connected conceptual network. The main challenges of this problem space include: 1) linking quality (correctly identifying which terms to link and which entry to link to with minimal effort on the part of users), 2) efficiency and scalability, and 3) generalization to multiple knowledge bases and web-based information environment. We present the NNexus approach that utilizes subject classification and other metadata to address these challenges. We also present evaluation results demonstrating the effectiveness and efficiency of the approach and discuss ongoing and future directions of research.