EuroMISE centrum – Cardio, co-holder organisation University of Economics, Prague

Project Stepper – Overview

There are two main goals of the Stepper project:

Methodology of step-by-step approach tries to find new way how to transform text documents into their formal representation. Most of similar approaches assume extensive interaction between domain expert and the knowledge engineer which fluently leads to final formal model. This one-step formalization process, however brilliant its output may be, has two critical defects. First is lack of transparency and the second is mixture of expertise. We could hardly find transparent way from part of source text to its adequate fragment in formal model if there is only one large formalization step. And if we look at part of knowledge in formal model we won’t be sure whether its origin lays in source text or if it’s only subjective interpretation of domain expert (or even worse totally external knowledge given by domain expert). Both of these problems are solved in methodology of step-by-step approach to text document formalization. Main idea of step-by-step approach is to divide formalization process into multiple steps where result of each step is new version of source document (new level) in deeper state of formalization. All knowledge elements in neighbor levels are interconnected with direct links so it’s easy to find ancestors of any knowledge element at any level. That could be helpful both in separating the different expertise, as well as in making the process more transparent and allowing for easier verification and correction of misinterpretations. Methodology is based on markup language XML and other XML tools.

Very shortly we find out that the methodology of Step-by-step approach itself is useless without software support. Hence we started development of application Stepper, which is a kind of XML editor including many features specialized on step-by-step approach.

We were testing the methodology in medical domain, namely in formalization of WHO/ISH hypertension guidelines. As a result we prepare practical implementation called Guide-X.

Development of Stepper shows us, that we need some way how to explicitly expressed transformations of knowledge in each step. Well known standard for transformation XSLT doesn´t solve all of our problems, hence we create our own transformation language based on XML called XKBT – (XML knowledge block transformation).

Within the framework of this project also several papers were published.

Date: September 21, 2017