Using Data Mining Techniques for Exploring the key Features of Plant Dynamics Upon a Newly Built Plant Trait Database
Michael Städler, Renée M. Bekker, Jens Finke, Dierk Kunzmann, Michael Sonnenschein · 2005
Most data mining techniques have rarely been used in ecology. To address the specific needs of scientists analysing data from a plant trait database developed during the LEDA project, a web-based data mining tool has been devel-oped. This paper presents the DIONE data miner and the project it has been developed in. It addresses the nature of plant trait data from a data mining perspective and points out problems that arise when preparing data for being mined. Furthermore it contains information about data mining results on plant trait data achieved so far and conclu-sions towards the applicability of data mining to the fields of ecology, and the newly emerging field of research re-cently addressed as ecoinformatics. 1. The LEDA Project The LEDA project (Life-history traits of the Northwest European Flora- a database), funded by the Euro-pean union (contract no. EVR1-CT-2002-40022) currently builds up a European-wide database with in-formation on 30 plant characteristics for about 3000 plant species of Northwest Europe. Data comes from scattered database initiatives, from literature and from new measurements and observations conducted during the project and is collected in the so called LEDA Traitbase [Knevel 2003]. For the first time in his-tory, plant trait data covering a substantial part of a continent will be available from a single database ac-cessible through the WWW. The LEDA Traitbase carries the potential of generating a huge amount of knowledge on species group-ing, assembly rules of communities and the understanding of species ’ mobility through the landscape re-sulting in explaining invasiveness or endangerment of species. Its web interface consists of a portal appli-cation and sub-applications for the purposes of online data input, uploading data, performing online que-ries, online reviewal of data [Ahlers 2005], administrative tasks, editorial tasks, and for the purpose of data mining. The web-applications will be publicly accessible through the LEDA web portal