In recent years, the process humans adopt to learn a foreign language has moved from the strict "Grammar –Translation" method, which is based mainly on grammar and syntax rules, to more innovative processes, resulting to the more modern "Communicative approach". As its name states, this approach focuses on the coherent communication with native speakers and the cultivation of oral skills, without taking into consideration, at least at the first stages, the rules that govern the language. The same trend seems to have been applied to the way machinery can be "educated" to comprehend and reproduce the unfamiliar, human language. The "rule based" Natural Language Generation (NLG) and Natural Language Understanding (NLU) algorithms, on one hand, and the "text based" Large Language Models (LLMs), on the other, are two, analogous to the two human foreign language learning processes, subareas of Natural Language Processing (NLP). This paper presents these two alternative approaches, LLMs (a technology having surfaced as an influential catalyst of NLP, during last years) on the one hand and NLG/NLU on the other, highlighting their applications, their technologies, their capabilities, their differences, their strengths and weaknesses and the challenges they present, contributing to a deeper comprehension of the evolving landscape of Artificial Intelligence and human-computer communication.

Large Language Models (LLMs) have been the cutting-edge technology in natural language processing (NLP) in recent years, making machine-generated text indistinguishable from human-generated text. On the other hand, “rule-based” Natural Language Generation (NLG) and Natural Language Understanding (NLU) algorithms were developed in earlier years, and they have performed well in certain areas of Natural Language Processing (NLP). Today, an arduous task that arises is how to estimate the quality of the produced text. This process depends on the aspects of text that you need to assess, varying from correct grammar and syntax to more intriguing aspects such as coherence and semantical fluency. Although the performance of LLMs is high, the challenge is whether LLMs can cooperate with rule-based NLG/NLU technology by leveraging their assets to overcome LLMs’ weak points. This paper presents the basics of these two families of technologies and the applications, strengths, and weaknesses of each approach, analyzes the different ways of evaluating a machine-generated text, and, lastly, focuses on a first-level approach of possible combinations of these two approaches to enhance performance in specific tasks.

The article describes the academic data, which derived from a University E-government analytic platform, which supports the facilitation of blended learning in a Greek University during and after the COVID19 outbreak [1,2]. University e-government services refer to a set of information systems that facilitate the functionalities of the University and enable the management of the underlying information [3,4]. These educational, research and managerial services, also called U-EGOV, follow the four stages of e-government (Presence, Interaction, Transaction, Transformation) [5]. In the presented approach, the data was aggregated from the university services with an automated process and includes all the individual U-EGOV services, that is the synchronous and asynchronous educational platforms, the teleconferencing tool, etc. The dataset created contains information about the courses, the assignments, the grades, the examinations, as well as other significant academic elements of the synchronous and the asynchronous education that takes place in the University. The analysis spans from the spring semester of the academic year 2019–2020, the winter semester of the academic year 2020–2021 to the spring semester of 2020–2021 (three semesters in total). The sample consists of 4800 records and after the preprocessing 4765 records (statistics of courses attended by students) which include 1661 unique students within the university in twenty (20) courses. We have followed an educational data mining approach on the collected data by utilizing an automated data aggregation mechanism to gather data for the selected courses, in order to enhance the learning process and the quality of services. The dataset can be reused: i) as a reference point to measure the quality of the academic outputs and its progress through the years and ii) as a basis for similar analysis in other Higher Educational Institutions (HEIs).

Aggression in patients with dementia poses significant caregiving and clinical issues. In this work, fusion approaches—Early Fusion and Late Fusion—were compared to classify aggression using audio and visual signals. Early Fusion integrates the extracted features of the two modalities into one dataset before classification, while Late Fusion integrates the prediction probabilities of standalone audio and visual classifiers with a meta-classifier. Both models were tested using a Random Forest classifier with five-fold cross-validation, and the performance was compared on accuracy, precision, recall, F1-score, ROC-AUC, and inference time. The results showcase that Late Fusion is superior to Early Fusion in terms of accuracy (0.876 vs. 0.828), recall (0.914 vs. 0.818), F1-score (0.867 vs. 0.835), and ROC-AUC score (0.970 vs. 0.922), proving more suitable for high-sensitivity use cases like healthcare and security. However, Early Fusion exhibited higher precision (0.852 vs. 0.824), indicating that in cases when false positives are a requirement, Early Fusion is preferable. Paired t-tests were applied for statistical comparison and indicate that precision alone is significantly different, with the advantage of Early Fusion. Late Fusion also performs slightly less in inference time, which makes it suitable for use in real-time systems. These findings provide significant information on multimodal fusion strategies and their applicability in the detection of aggressive behavior, which can contribute to the development of efficient monitoring systems for dementia care.

Nowadays, the business area cannot be sustainable and efficient without the presence of the Supply Chain. However, Supply Chain Management is by no means an easy task. Experts, in their effort to achieve the most efficient Supply Chain Management, have turned their attention to the management of knowledge related to supply chains. Thus, they model the concepts and the semantic relationships between them in Supply Chain Networks and create conceptual models and ontologies. In this paper, a survey of the existing ontologies in this field is carried out, with the aim of creating a new ontology of the Supply Chain that will unify its structural elements and lead to the integration of all supply systems. For this purpose, 22 ontological models from different supply systems, from over 90 sources, have been collected, briefly presented and commented on. These models, although being an intersection in the effort to model business operations and delineate a good basis for businesses to engage in Logistics processes, they lack an adequate formulation of appropriate semantics and terminology to describe all the different functions of the Supply Chain. Consequently, this lack inevitably leads to incompatible interpretations and uses of the knowledge that results from intercompany transactions.

This paper introduces a classification model that detects and classifies argumentative behaviors between two individuals by utilizing a machine learning application, based on the MediaPipe Holistic model. The approach involves the distinction between two different classes based on the behavior of two individuals, argumentative and non-argumentative behaviors, corresponding to verbal argumentative behavior. By using a dataset extracted from video frames of hand gestures, body stance and facial expression, and by using their corresponding landmarks, three different classification models were trained and evaluated. The results indicate that Random Forest Classifier outperformed the other two by classifying argumentative behaviors with 68.07% accuracy and non-argumentative behaviors with 94.18% accuracy, correspondingly. Thus, there is future scope for advancing this classification model to a prediction model, with the aim of predicting aggressive behavior in patients suffering with dementia before their onset.