Klasykacja operatorów metatekstowych i częstość ich występowania w krótkich tekstach naukowych w języku polskim

Abstract

The article presents an analysis of the usage frequency of different types of metatext markers in short scientific texts written in Polish. A well-known classification by Hyland (1998, 2005) was used with additional binary classifications by Bunton (1999) and Dahl (2004). Data mining was performed on the data using rule-generating algorithm OneRule, decision tree J48, bayesian Naive Bayes Classifier and k-Neares Neighbour classifier, in order to analyse relations between the classes of metatext markers found in the texts. The outcomes of the analysis may be used to simplify classification of metatext markers. Information on metatext markers classes frequency may also be used for preparing or adapting texts in research on the influence of metatext markers on reading and, eventually, for automatic text structure analysis and abstract generation.