Skip to Content


warning: Creating default object from empty value in /home/goodoldai/public_html/modules/taxonomy/ on line 34.

MDE of Rule-based SOA (rBPMN)

Posted in

Rule-based Business Process Modeling Notation (rBPMN)

Following the principles of MDE, we integrate a new rule gateway type into BPMN business process models on the metamodel level. The result is a language called rBPMN (Rule-based BPMN) that facilitates business process modeling by domain experts and enables transformations of such process models into different SOA implementation platforms. This high-level modeling approach allows developers to focus on a problem domain rather than on an implementation technology.

Business processes are represented by business process models. This requires a notation that provides notational elements for the conceptual elements of process metamodels. The rBPMN process notation is associated with the rBPMN process metamodel level and with the rBPMN process model level (by using the MDE approach). Each rBPMN process model is expressed in the rBPMN process notation associated with the rBPMN process metamodel.

Policy Modeling Language

PML is a language aiming at facilitating the process of defining policies. PML helps policy designers by abstracting away the low-level details of various existing policy languages and by enabling high-level specification of policies using a graphical tool. The tool will support the transformation of high-level policy models to the low-level policy language of the policy designer's choice.

PML and other Policy Languages

Modeling Online Presence

Posted in

The objective of Modeling Online Presence is to enable the integration and exchange of data related to users' online presence. Someone's online presence reflects her/his current use of and interactions with online sevices. It includes elements of the temporary state of the user's interactions with the services. The project provides a Semantic Web online presence ontology (OPO) for representing rich data about a user's online presence in RDF. The rationale is as follows.

RAS - Reasoning in Assessment Systems

Posted in

The RAS project focuses on intelligent analysis of students’ knowledge in QTI-FAR systems, using description logics (DLs) reasoning techniques. Specifically, the objective of RAS is to support semantic analysis and evaluation of students’ answers and solutions acquired through the use of a QTI-FAR system. We assume that the students’ answers are submitted to the resoning machine as OWL models, transformed from a QTI model. Ontology model conforms to OWL metamodel, which in turn conforms to the  Ontology Definition Metamodel. In processing an open-ended question, the reasoner deploys these transformations to generate the tableau model equivalent to the QTI model of the student’s answer. As a consequence, if two different students submit two syntactically different but semantically correct answers to an open-ended question, the reasoner will detect that both answers are correct (satisfiable).

Project team: Nenad Krdzavac, Sonja Radenkovic, and Vladan Devedzic


Posted in


Synesketch is a Web's first free open-source software library for textual emotion recognition and visualization – code that feels the words visually.

It is a form of digital synesthesia, a bridge between words, emotions, and images. While one types, Synesketch dynamically transfers the text into animated visual patterns. Emotional content of the text is represented via various Processing graphics (everyone can build her/his own). Our project is a result of a research that spreads out through several diverse fields – from natural language processing techniques based on WordNet, across Ekman's research of emotions, to color psychology, visual design, data visualizations, and affective computing. 

Besides injoying our visual artworks, you are welcome to create your own Synesketch visualizations, or build serious emotion sensing or synesthesia-like applications, with Java and Processing. Enhancing human communication, expression, and inspiration is the main goal of the project. 

Synesketch Graphics

QTI-FAR - A Framework and ARchitecture for Assessment Systems

Posted in

This project is about applying intelligent reasoning techniques in e-assessment systems. We have developed QTI-FAR - a framework for design and implementation of e-assessment systems based on the IMS Question and Test Interoperability (QTI) standard.  We call such systems QTI assessment systems, or just QTI systems. The framework is developed starting from the OMG Model Driven Architecture (MDA) software engineering standards. It includes a QTI metamodel (in UML and ECore) and model transformations for QTI assessment systems, as well as a QTI model repository that could be reused in implementation of other QTI models that an assessment expert may suggest in a specific assessment system. It also specifies an architecture for QTI assessment systems that is reusable, extensible, and facilitates interoperability between its component systems. QTI-FAR systems are QTI assessment systems implemented using QTI-FAR. A use case model for a simple choice example question in QTI-FAR system is shown here. Details about the DL reasoner used by this QTI-FAR system are shown in the RAS project.
Project team: Sonja Radenkovic, Nenad Krdzavac, Vladan Devedzic

The Neuroph Project

Posted in

Neuroph is a lightweight Java framework for developing neural networks. It can be used to create and train common neural network architectures.


It has small number of basic classes that correspond to the basic neural network concepts, so it is very intuitive and easy to learn. If you are a beginner in neural networks, or you just want to try how they work without going into complicated theory and implementation, or you need them for your research project - the Neuroph is a good choice for you.

Find out more about Neuroph at, or contact Zoran Sevarac, the Neuroph founder and developer.


The objective of this project is to create easy to use, flexible and well documented development environment for neural networks . The Neuroph framework provides a set of Java classes which can be used to easily create neural networks in Java code, and also the GUI application to create, train and save neural networks as Java components. Created networks can then be used in Java programs.

Basic Info


Syndicate content