Building Knowledge Bases from the Web

  • Rajeev Rastogi | Amazone, Bangalore

The web is a vast repository of human knowledge. Extracting structured data from web pages can enable applications like comparison shopping, and lead to improved ranking and rendering of search results. In this talk, I will describe a technique that exploits machine learning models to automatically extract records from web pages without human supervision. Specifically, we use Markov Logic Networks (MLNs) to capture content and structural features in a single unified framework, and devise a fast graph-based approach for MLN inference.

Watch Next