Search

This paper introduces the USPDATRO dataset. This is a speech dataset, in the Romanian language, constructed from open data, focusing on under-represented voice types (children, young and old people, and female voices). The paper covers the methodology behind the dataset construction, specific details regarding the dataset, and evaluation of existing Romanian Automatic Speech Recognition (ASR) systems, with different architectures. Results indicate that more under-represented speech content is needed in the training of ASR systems. Our approach can be extended to other low-resourced languages, as long as open data are available.

In modern societies, digitalization plays a tremendously important role for people and businesses. Apart from an economic representation, competitiveness characterizes a society from political, cultural, or human points of view. In this article, we aim to highlight the role of digital development from a competitiveness perspective, as there are few studies related to this relationship. The empirical investigation is based on panel data analysis for European Union countries for 2017–2022, considering the digital economy and society index (DESI) and the index developed by International Institute for Management Development (IMD), respectively IMD world competitiveness index. The results obtained are reported both for general indices and for the components of DESI, presented separately for the groups of Central and Eastern European countries and Western European countries. They indicate different influences for the two groups of countries, with only a few common aspects. The most obvious is the case of skilled labor. This aspect demonstrates the link between the various dimensions of digitalization and changes in human capital development strategies, as they appear in the specialized literature.

Shear-wave elastography (SWE) showed the absence or presence of significant differences among stable kidney allograft function and allograft dysfunction. We evaluated the variability of kidney allograft stiffness in relation to allograft dysfunction, respectively, in terms of a correlation of stiffness with patients’ characteristics. A single-center prospective study on patients who had undergone renal transplantation was conducted between October 2017 and November 2018. Patients were clinically classified as having a stable allograft function or allograft dysfunction. SWE examinations performed by the same radiologist with a LOGIQ E9 were evaluated. Ten measurements were done for Young’s modulus (kPa) at the level of allograft cortex and another ten at the level of medulla. Eighty-three SWE examinations from 63 patients, 69 stable allografts, and 14 allografts with dysfunction were included in the analysis. The intra-examinations stiffness showed high variability, with the quantile covariation coefficient ranging from 2.21% to 45.04%. The inter-examinations stiffness showed heterogeneity (from 28.66% to 42.38%). The kidney allograft cortex stiffness showed significantly higher values in cases with dysfunction (median = 28.70 kPa, interquartile range (IQR) = (25.68−31.98) kPa) as compared to those with stable function (median = 20.99 kPa, interquartile range = (16.08−27.68) kPa; p-value = 0.0142). Allograft tissue stiffness (both cortex and medulla) was significantly negatively correlated with body mass index (−0.44, p-value < 0.0001 for allograft cortex and −0.42, p-value = 0.0001 for allograft medulla), and positively correlated with Proteinuria/Creatinuria ratio (0.33, p-value = 0.0021 for allograft cortex and 0.28, p-value = 0.0105 for allograft medulla) but remained statistically significant only in cases with stable function. The cortical tissue stiffness proved significantly higher values for patients with allograft dysfunction as compared to patients with stable function, but to evolve as an additional tool for the evaluation of patients with a kidney transplant and to change the clinical practice, more extensive studies are needed.

The benefits of using comparable corpora for improving translation quality for statistical machine translators have been already shown by various researchers. The usual approach is starting with a baseline system, trained on out-of-domain parallel corpora, followed by its adaptation to the domain in which new translations are needed. The adaptation to a new domain, especially for a narrow one, is based on data extracted from comparable corpora from the new domain or from an as close as possible one. This article reports on a slightly different approach: building an SMT system entirely from comparable data for the domain of interest. Certainly, the approach is feasible if the comparable corpora are large enough to extract SMT useful data in sufficient quantities for a reliable training. The more comparable corpora, the better the results are. Wikipedia is definitely a very good candidate for such an experiment. We report on mass experiments showing significant improvements over a baseline system built from highly similar (almost parallel) text fragments extracted from Wikipedia. The improvements, statistically significant, are related to what we call the level of translational similarity between extracted pairs of sentences. The experiments were performed for three language pairs: Spanish-English, German-English and Romanian-English, based on sentence pairs extracted from the entire dumps of Wikipedia as of December 2012. Our experiments and comparison with similar work show that adding indiscriminately more data to a training corpus is not necessarily a good thing in SMT.