Categorization of schematic organized information by means of ROC curves.

The information organized in knowledge schemes (scripts) needs to be studiet by techniques in which elements are categorized as test material: Letters, syllables, words, sentences, pictures, etcetera. One of these techniques considers subjective scaling of the stimulus by means of rating scales (v.g. Torgerson, 1958). Furthemore, the application of rating scales has been a very useful technique applied in signal detection theory (SDT). With this technique it is possible to get the receiver operating characteristic or ROC curve. That curves represents ordinal data. This technique constitutes a natural categorization for estimations of ROC curves. On the other had, the presentation of one model of categorization applied to events included in the scheme. The categorization represents a first level to escale the information. A second level is the suitable application of indices of sensitivity. The third level is the statistical modelling. We have carried out initial research at the first an second level (v.g. Pelegrina, Salvador y Bull, 1999). In the present work we are analized possibilities of modelling when the data are considered or analiyzed by means of ROC analysis. This technique has been largely used in biomedical research(radiology)unlike psychology(v.g. Zou y Hall, 1997). However, this is a model of data representation complementary between the scheme theory and the SDT.