Large-Scale Knowledge Acquisition from Botanical Texts

Free text botanical descriptions contained in printed floras can provide a wealth of valuable scientific information. In spite of this richness, these texts have seldom been analyzed on a large scale using NLP techniques. To fill this gap, we describe how we managed to extract a set of terminological resources by parsing a large corpus of botanical texts. The tools and techniques used are presented as well as the rationale for favoring a deep parsing approach coupled with error mining methods over a simple pattern matching approach.

• Type: Conference
• Year: 2007
• Where: NLDB
• Authors: François Role, Milagros Fernandez Gavilanes, Eric Villemonte de la Clergerie
• Keywords: Deep Parsing Approach, Information System, NLDB 2007, Text Botanical Descriptions, Valuable Scientific Information