Intelligent Linguistic System for the Grammar of the Romanian Language
Ioan Florin Cătălin Nitu, Traian Rebedea
Stress, frustration, boredom, and fatigue in online engineering education during the pandemic
Valentina Iuliana Manea, Tudor Macavei, Costin Pribeanu
1 University Politehnica of Bucharest
313 Splaiul Independentei, Bucharest, Romania
2 Open Gov SRL
95 Blvd. Alexandru Ioan Cuza, Bucharest, Romania
Abstract. The field of natural language processing is not as strongly developed for the Romanian language as it is for others, as is the English language. Writing texts correctly has always been a necessity, and the development of tools that will be useful in this need is critical. The proposed correction system receives a sentence with grammatical errors and corrects it, using state-of-the-art technologies to perform this operation such as attention-based neural models based on Encoder-Decoder Transformers. These are a cornerstone in the development of intelligent tools for processing – translating, summarizing, or proofreading – texts and are the foundation for this project. The paper uses RONACC, the first corpus for grammatical corrections in Romanian for modeling, training, testing, and validating the project. Using a very large dataset with over a million learning examples, an average BLEU score of 45.29 points was obtained, in a rather short training time executed on several GPUs. However, even a smaller dataset of only fifty thousand examples with as many as one hundred epochs achieves an average BLEU score of 33.29 points in three hours.
Keywords: Romanian language, grammar, transformers, attention, encoder-decoder
Cite this paper as:
Nitu, I. F. C., Rebedea, T. Intelligent Linguistic System for the Grammar of the Romanian Language. International Journal of User-System Interaction 13(4), 183-198, 2020.
1 Technical University of Civil Engineering
Bd. Lacul Tei 122-124, Bucharest, Romania
2 Academy of Romanian Scientists
Splaiul Indpendentei 54, Bucharest, Romania
Abstract. The educational crisis process generated by the Coronavirus COVID-19 had major consequences on the academic and personal life of university students. Universities had to move from the traditional classes and laboratories to the online platform. As extant research shows, this shift to online education had many negative consequences. The objective of this research is to analyze in more detail the relationship between stress, frustration, boredom, and fatigue and to identify the main factors that affect the online learning experience of students during the pandemic. The analysis is a mix of a quantitative and qualitative study that has been done on a sample of 177 university students from a university of construction engineering in Bucharest. The analysis of quantitative data revealed a significant and relatively high correlation between stress, frustration, boredom, and fatigue. The qualitative analysis sheds light on the relationship between these variables and the loss of attention, learning difficulties, and lack of face-to-face communication, socialization, and interaction.
Keywords: online lectures, frustration, stress, boredom, pandemic, engineering education
Cite this paper as:
Manea, V. I., Macavei, T., Pribeanu, C. Stress, frustration, boredom, and fatigue in online engineering education during the pandemic. International Journal of User-System Interaction 13(4), 199-214, 2020.