The national e-health systems implemented in Serbia as ”MojDoktor” (MyDoctor) and in Macedonia as ”MojTermin” (MyAppointment) are based on the same integrated health information platform. Its architecture was designed to allow for process oriented development with agile methodologies. This methodology allowed for fast deployment and adoption, but a change in the architecture to a more formal approach is required to assure its extensibility, soundness, interoperability and standardization. In this paper, we propose a formalization of the system design and its implementation as a Model Driven Architecture®. We develop a set of formal models on several abstraction levels and explain how different layers of the MDA framework are covered using them. We also propose a model for data transformations that will provide interfaces for interoperability between an external and our system.

Language is a unique ability of human beings. Although relatively simple for humans, the ability to understand human language is a highly complex task for machines. For a machine to learn a particular language, it must understand not only the words and rules used in a particular language, but also the context of sentences and the meaning that words take on in a particular context. In the experimental development we present in this paper, the goal was the development of the language model SRBerta—a language model designed to understand the formal language of Serbian legal documents. SRBerta is the first of its kind since it has been trained using Cyrillic legal texts contained within a dataset created specifically for this purpose. The main goal of SRBerta network development was to understand the formal language of Serbian legislation. The training process was carried out using minimal resources (single NVIDIA Quadro RTX 5000 GPU) and performed in two phases—base model training and fine-tuning. We will present the structure of the model, the structure of the training datasets, the training process, and the evaluation results. Further, we will explain the accuracy metric used in our case and demonstrate that SRBerta achieves a high level of accuracy for the task of masked language modeling in Serbian Cyrillic legal texts. Finally, SRBerta model and training datasets are publicly available for scientific and commercial purposes.

Government institutions have released a large number of datasets on their open data portals, which are in line with the data transparency and open government initiatives. With the purpose of making it more accessible and visible, these portals categorize datasets based on different criteria like publishers, categories, formats, and descriptions. However, some of this information is often missing, making it impossible to find datasets in all of these ways. As a result, with the number of datasets growing further on the portals, it is getting harder to obtain the desired information. This paper addresses this issue by introducing EODClassifier framework that suggests the best match for the category where a dataset should belong to. It relies on formal concept analysis as a means to generate a data structure that will reveal shared conceptualization originating from tags' usage and utilize it as a knowledge base to categorize uncategorized open datasets.

Producing a new high-quality text corpus is a big challenge due to the required complexity and labor expenses. High-quality datasets, considered a prerequisite for many supervised machine learning algorithms, are often only available in very limited quantities. This in turn limits the capabilities of many advanced technologies when used in a specific field of research and development. This is also the case for the Serbian language, which is considered low-resourced in digitized language resources. In this paper, we address this issue for the Serbian language through a novel approach for generating high-quality text corpora by improving text recognition accuracy for scanned documents belonging to Serbian legal heritage. Our approach integrates three different components to provide high-quality results: a BERT-based large language model built specifically for Serbian legal texts, a high-quality open-source optical character recognition (OCR) model, and a word-level similarity measure for Serbian Cyrillic developed for this research and used for generating necessary correction suggestions. This approach was evaluated manually using scanned legal documents sampled from three different epochs between the years 1970 and 2002 with more than 14,500 test cases. We demonstrate that our approach can correct up to 88% of terms inaccurately extracted by the OCR model in the case of Serbian legal texts.

Electric power supply companies increasingly rely on enterprise IT systems to provide them with a comprehensive view of the state of the distribution network. Within a utility-wide network, enterprise IT systems collect data from various metering devices. Such data can be effectively used for the prediction of power supply network vulnerability. The purpose of this paper is to present the Enterprise Service Bus (ESB)-based Sensor Web integration solution that we have developed with the purpose of enabling prediction of power supply network vulnerability, in terms of a prediction of defect probability for a particular network element. We will give an example of its usage and demonstrate our vulnerability prediction model on data collected from two different power supply companies. The proposed solution is an extension of the GinisSense Sensor Web-based architecture for collecting, processing, analyzing, decision making and alerting based on the data received from heterogeneous data sources. In this case, GinisSense has been upgraded to be capable of operating in an ESB environment and combine Sensor Web and GIS technologies to enable prediction of electric power supply system vulnerability. Aside from electrical values, the proposed solution gathers ambient values from additional sensors installed in the existing power supply network infrastructure. GinisSense aggregates gathered data according to an adapted Omnibus data fusion model and applies decision-making logic on the aggregated data. Detected vulnerabilities are visualized to end-users through means of a specialized Web GIS application.