Web-annotations for humans and machines

Abstract

We propose to manually annotate web pages with computer-processable controlled natural language. These annotations have well-defined formal properties and can be used as query relevant summaries to automatically answer questions expressed in controlled natural language, and as the basis for other forms of automated reasoning. Last, but not least, the annotations can also serve as human-readable summaries of the contents of the web pages. Arguably, annotations written in controlled natural language can bridge the gap between informal and formal notations and leverage true collaboration between humans and machines. This is a position paper that proposes a solution combining existing methods and techniques to achieve a highly relevant practical goal, namely how to effectively access information on the web. However, our solution introduces a "chicken and egg" problem: a critical mass of web annotations will be necessary that people perceive the value of these annotations and start annotating web pages themselves. Only the future will show whether this - basically non-technical - problem can be solved. © Springer-Verlag Berlin Heidelberg 2007.