Making Biographical Data in Wikipedia Readable: A Pattern-based Multilingual Approach

摘要

In this paper we present Biografix, a pattern based tool that simplifies parenthetical structures with biographical information, whose aim is to create simple, readable and accessible sentences. To that end, we analysed the parenthetical structures that appear in the first paragraph of the Basque Wikipedia, and concentrated on biographies. Although it has been designed and developed for Basque we adapted it and evaluated with other live languages. We also perform an extrinsic evaluation with a question generation system to see if Biografix improve its results.