Organizing and searching the world wide web of facts -- step two

Abstract

As part of a large eort to acquire large repositories of facts from unstructured text on the Web, a seed-based frame- work for textual information extraction allows for weakly supervised extraction of class attributes (e.g., side ee cts and generic equivalent for drugs) from anonymized query logs. The extraction is guided by a small set of seed at- tributes, without any need for handcrafted extraction pat- terns or further domain-specic knowledge. The attributes of classes pertaining to various domains of interest to Web search users have accuracy levels signican tly exceeding cur- rent state of the art. Inherently noisy search queries are shown to be a highly valuable, albeit unexplored, resource for Web-based information extraction, in particular for the task of class attribute extraction.