Implementation of method for generating concept map from unstructured text in the Croatian language

Abstract

This paper describes implementation and evaluation of a method for automatic creation of concept map from unstructured text written in the Croatian language. The method combines statistical and data mining techniques with domain thesaurus and tools specific for linguistic analysis of the Croatian language. We created concept maps from legal documents in the field of science, education and technology and evaluated the content of created concept maps comparing them with manually created summaries. Obtained results confirmed that using the described method, it is possible to create concept maps from documents in a specific area, and those maps illustrate the contents of the source documents well.