Computational Science and Its Applications – ICCSA 2023

Automated testing and evaluation of interfaces is a well-established reality supported by many tools that shorten the time to deploy new software versions to the user. However, exploring users’ emotions while interacting with interfaces as a tool to further increase the quality of traditional usability evaluation methods is still far from being a reality. This work uses the automatic analysis of users’ emotions while interacting with touchable interactive urban devices to detect usability issues. To this end, a coupled approach is implemented: the data is acquired from the interaction, and user emotions are extracted and processed to determine the emotional status during the interaction. This data is integrated into a web application so that designers can further improve the quality of the interface in the presence of negative emotions. Results show that the experimental tests showed that different users manifest similar negative emotions in the same contexts, which is a clear sign of usability issues that are to be corrected by the design team.

Deep Learning has become a popular tool for addressing complex tasks in many computer vision applications. Label diffusion methods have also been a very effective technique for getting accurate segmentations of real-world images, as they combine user autonomy, versatility and accurateness through a user-friendly interface. In this paper, we propose a seeded segmentation framework for partitioning real-world images by combining deep contour learning and graph-based label propagation models. More precisely, our approach takes a CNN-type contour detection network to learn graph edge weights, which are used as input to solve a coupled energy minimization problem that diffuses the user-selected annotations to the desired targets. To accurately extract deep features from image contours while generating diffusion maps, we train a deep learning architecture that integrates a hierarchical neural network, a graph-based label propagation model and a loss function, allowing the coupled training mechanism to refine the results until convergence. We attest to the effectiveness and accuracy of the proposed approach by conducting both quantitative and qualitative assessments with existing seeded image segmentation methods.

Even though significant progress has been made in the field of pattern recognition in recent years, handwritten digit recognition remains an exciting challenge. Due to its widely possible application, this issue has attracted much attention. Nowadays, different methods are available for handwritten digit recognition. The current demand is for researchers to find new techniques for better performing handwritten digit recognition problems. Therefore, this paper aims to propose a new approach, the Siamese network with Gabor filter, for handwritten digit recognition. Inspired by several studies on the Siamese network and Gabor filter separately, which have already achieved superb performances, this research will bring out the best qualities of their fusion. The computational experiments have been conducted on a handwritten digit image of the MNIST dataset. Empirically, the results implied that the proposed Siamese network with the Gabor filter algorithm outperformed the classical Siamese network and other existing methods in terms of accuracy.

Time series data is of crucial importance in different domains, such as financial and medical applications. However, obtaining a large amount of labeled time series data is an expensive and time-consuming task, which becomes the process of building an effective machine learning model a challenge. In these scenarios, algorithms that can deal with reduced amounts of labeled data emerge. One example is Semi-Supervised Learning (SSL), which has the capability of exploring both labeled and unlabeled data for tasks such as classification. In this work, a kNN graph-based transductive SSL approach is used for time series classification. A feature extraction step, based on imaging time series and obtaining features using deep neural networks is performed before the classification step. An extensive evaluation is conducted over four datasets, and a parametric analysis of the nearest neighbors is performed. Also, a statistical analysis over the obtained distances is conducted. Results suggest that our methods are suitable for classification and competitive with supervised baselines in some datasets.

Among the geometric bodies of revolution, the torus of revolution stands out, which can be generated from a circumference, lemniscate curves and the figure-of-eight curve. Given the classical definition used in Mathematics, interest arises in finding other curves that generate the torus of revolution when rotating around an axis. In this article, carrying out the respective analyzes and the necessary programming using the Mathematica software, allowed us to carry out the necessary calculations and geometric visualizations of the mathematical object: This is how a torus of revolution was built from epicycloid curves in their parametric form. The study was extended by determining curves that were on the torus generated by epicycloid curves, which when properly projected to planes, curves that present beautiful symmetries were obtained. When the points of these curves are taken correctly, special irregular polygons are obtained.With the obtaining of these results, a satisfactory answer to the research question was obtained, as well as a way of defining it. In addition, it has shown us a wide path of research on the different curves that a torus of revolution can generate.

Labor potential estimation is a complex analysis task of a large data sets of digital profiles of workers. Digital profile is a structure that contains heterogeneous data and has high complexity. The attribute space of digital profiles datasets has a large dimension. Individual features are distinguished by a large variability of types and include qualitative and quantitative characteristics that can be both continuous or discrete. This fact significantly complicates the use of traditional statistical methods and hinder manual selection of machine learning algorithms for labor potential estimation. Automated machine learning (AutoML) can help to deal with this problem. The results of AutoML can significantly depend on the used AutoML library. AutoML libraries differ in the set of algorithms among which they select the best one, in the constrains that are used to stop the search of the optimal solution, in the techniques used for hyperparameters optimization and in metrics that are calculated for the obtained results. In this work we develop an AutoML framework that unites several commonly used AutoML libraries and enables use of common constrains and metrics. Processing dataset about employees’ performance using the developed framework we show the applicability of the AutoML for the task of labor potential estimation.

The presented paper is devoted to the numerical study of the applicability of 1D seismic inversion and 2D machine learning based inversion for fracture model reconstruction. Seismic inversion is used to predict reservoir properties. Standard version is based on a one-dimensional convolutional model, but real geological media are more complex, and therefore it is necessary to determine conditions where seismic inversion gives acceptable results. For this purpose, the work carries out a comparative analysis of one-dimensional and two-dimensional convolutional modeling. Also, machine learning methods have been adopted for 2D fracture model reconstruction. We use UNet architecture and 2D convolutional model to create a training dataset. We perform numerical experiments for a realistic synthetic model from Eastern Siberia and Sigsbee model.

The paper presents a simple and robust approach for calculating frequency-dependent rays in three-dimensional media. The proposed ray tracing procedure simulates propagation of locally plane fragment of a wave front. Ray properties depend on velocity distribution in some sub-volume around the ray and on wavelength at each point. We provide a numerical comparison of “fat” rays approach with the “exact” Helmholtz solver using complex 3D SEG salt model. This comparison show promise of using the concept of “fat” rays in 3D seismic data processing, i.e. comparison show that “fat” rays approach needs up to several orders less resources than Helmholtz solver, while the error between “fat” rays wave filed and “exact” solution does not exceed several percent.

Designing effective collaboration, visualization, and analysis of severe weather events (e.g., excessive rainfall accumulation, floods, severe thunderstorms, tornadoes, and droughts) for meteorologists on the web is challenging because it requires processing huge amounts of data from different sources, diverse technologies, as well as using appropriate visualization techniques. This paper presents an innovative web toolset for visualization, collaborative discussion, and analysis of severe weather events. It is composed of the tool Viewing events on Maps (VIMAPS), which allows users to visualize and analyze severe weather events using maps enriched with multivariate data glyphs to observe spatial relations from any desired angle; and the tool called Collaborative system for Meteorological Data Analysis (COMETA) that allows the collaborative analysis of severe weather events between remote users synchronously or asynchronously. These tools were tested using real data and showed that they fulfill the usability requirements of this class of application.

This paper investigates alternative clustering algorithms as part of a cluster-first, route-second method for solving a vehicle routing problem with drones. The problem consists of multiple purpose-built trucks, each having a single drone that can be launched to make deliveries, which set out to deliver products to customers. The drones are able to intercept and land on a moving truck after making a delivery, which speeds up the synchronisation and delivery time. Various different clustering algorithms are evaluated as part of a self-adaptive neighbourhood search differential evolution algorithm and applied to fifteen publicly available datasets, containing three types of customer distributions: double-centred, single-centred, and uniform. The customer nodes are segmented into clusters and each cluster is serviced by one truck and drone pair. Each cluster is solved as a travelling salesman problem with drone with interceptions using the nearest neighbour heuristic for initial solutions self-adaptive neighbourhood search differential evolution algorithm. A k-means clustering algorithm, the Gaussian mixture model, and agglomerative nesting are evaluated. From the study, it is conclusive that the Gaussian mixture model is the best clustering algorithm for double-centred datasets. The k-means algorithm is the best clustering algorithm for single-centred and uniform datasets. The contributions of this paper include the only cluster first-route second approach to solving the vehicle routing problem with drone with interceptions. This paper contributes to the computational sciences by using advanced computing capabilities for decision support to improve the performance of an intelligent transport system.

The article is devoted to the study of methodological features and justification of practical measures in the organization of strategic directions to ensure the financial stability of the transport enterprise. It is proved that financial analysis is a set of measures that are used in the process of studying the financial condition of a transport enterprise. The main indicators of the results of its business activities to identify summarize the weaknesses and strengths of the enterprise. Financial stability is a very important indicator of the real state of affairs of both the financial condition of the transport enterprise and the effectiveness of the cash flow management policy. Attention is focused on the fact that the level of financial stability of the enterprise directly depends on the chosen policy and strategy of financial stability management. Based on this, it should be said that the directions of minimizing the risks of reducing the financial stability of the enterprise are included in the system of financial controlling and management policy of the transport enterprise itself. It is established that to maintain the necessary level of financial stability of the insurance company in the transport industry and minimize the risk of reducing financial stability, it is important to have a quality management policy and the selected type and model of management of the main items that affect it, which include the level of current assets, reserves and sources of their formation. It is also necessary to research and allocate items in time to increase their own sources and provide them with material working capital. It is revealed that the optimal ratio of financial resources is also important. With the help of which it is possible to ensure an effective uninterrupted process of production and sale of products, goods and services due to their effective disposal. This, in turn, will contribute to the expansion and updating of the results of the company’s activities.

Under normal conditions, favorable connections with Slovenia and Austria allow residents of the Italian region of Friuli Venezia Giulia to take advantage of the different market conditions abroad. However, due to pandemic-issued border closures, the privilege of unbounded mobility was compromised. This contribution aims to understand the implications of a temporary border closure on the fuel retailing market by exploring quantitative data on fuel purchases made available by diverse public entities. In the first year of the pandemic, despite a national situation in which a strong contraction in sales of fuels was recorded, the Friuli Venezia Giulia region behaved significantly differently from Italy, especially in its border provinces of Gorizia and Trieste. The return to an open border has triggered strong international competition. As such, the results advance our knowledge of the impact of customer (im)mobility on the economy of a borderland.

The purpose of this research is to investigate the creation of financial resources in municipal budgets and to highlight significant concerns. Local self-government is crucial for the development of a robust economy in a nation, but without a balanced budget, a municipality cannot be independent or powerful. As a result, it's crucial to identify the mechanisms for optimization and assess the difficulties associated with creating local budgets. In the study, the sources of Georgian municipal budget creation are examined and discussed. The difficulties that now exist are highlighted based on the analysis of the study findings, and modifications that should be made are suggested based on the aforesaid facts. These changes will be crucial for the efficient operation of the local budgets.

Smart Cities represent the next big frontier for computational science. However, the real-world impacts of this transformation have been considerably slower than other domains of digital innovation. In this work, we study the interplay between the core properties of digital platforms and the urban innovation contexts that aim to promote digital transition as a means to generate value for cities and its citizens. The research methodology is based on a literature review, which aimed to characterise the key limitations preventing smart city initiatives from attaining the same level of fast paced innovation as other areas of computational science, and seek for alternative innovation practices. The results suggest that innovation practices in the context of smart city initiatives seem to be framed by a key trade-off between the idea that only global solutions may be able to capture the full benefits of digital innovation and the idea that each city is unique and must pave its own way towards digital transition. From the analysis of those results, we derive five design principles for new service-based platforms. These principles represent a new direction to address the specificities of smart cities and unleash the real-world impact of digital innovation in smart and sustainable cities.

In view of international trade transactions increase worldwide along with the Customs policy of Ukraine aimed at reducing government control and pressure over business, the issue of assigning the Authorized Economic Operator (AEO) status is becoming more and more important. Therefore, there is a strong need to develop a methodology for objective quantitative assessment of the qualitative characteristics of enterprises involved in international trade operations during Customs diagnostics in order to determine whether they comply with Customs laws to obtain the AEO status or not. A methodical approach to Customs diagnostics based on an integral indicator of compliance of an enterprise’s quality characteristics with Customs rules divided into five thematic blocks has been worked out, which allows determining an enterprise’s rating and avoid subjectivity in assigning a company the AEO status. To make a decision whether to assign a company the AEO status or not, it has been proposed to use a scale of compliance of an enterprise’s rating with one of the committee's conclusions such as unconditionally positive, positive, conditionally positive, negative, refusal to assign the status.

Since 2020, COVID-19 has raged throughout the world, and with recommendations to open windows and ventilate enclosed spaces, such as buses and trains, attention has been drawn to the issue of indoor environmental pollution. Therefore, it was hypothesized that measuring and reporting the air quality inside buses would allow passengers peace of mind. In this research, Internet of Things (IoT) sensors were installed inside buses, and the measured data were analyzed. With the cooperation of Sue Town in Japan, IoT sensors were installed in regional public transportation systems (i.e., Japanese Community Buses) and data that temperature, humidity, CO $$_2$$ 2 , noise, and pressure were collected. The measured data were displayed in real-time on signage inside the bus. As a preliminary step to predict the ability of the buses to maintain normal air quality inside the vehicles at all times, data accumulated on three buses over a six-month period were analyzed.

The paper considers the problem of spatial inequality of agglomerations. The paper proposes an approach to the estimation of spatial inequality based on the intermodal transport graph and data on the placement of urban services in the agglomeration cities. This approach has been experimentally tested on the agglomeration of St. Petersburg. The result is an assessment of the influence of cities in relation to each other, as well as the evaluation of each city by the availability of a particular type of service.

This paper examines a method for assessing spatial inequality through access to opportunities and urban services based on modeling an intermodal graph of accessibility of urban areas. The goal is to obtain ratings on the difference in physical access to services and opportunities on public transport by social groups. This is done by collecting data on the city’s residential areas and urban amenities that serve the functions of education, health, sports, and leisure. Calculation of travel time by public transport to the nearest service of each type is carried out using an intermodal graph. The result of the work is an assessment of the accessibility of services in terms of travel time, compared with urban planning standards.

The COVID-19 pandemic has recently shed light on the potential for social media as a means of spreading mis-, dis-, and malinformation. This paper investigates embedding and cluster-based topic modelling to characterise the COVID-19 infodemic on South African Twitter, which has largely remained unstudied during the COVID-19 pandemic. The best performing model is able to identify specific misinformation narratives, but these narratives are mostly found within more general topics. A more fine-grained model is trained, and is able to much better isolate rumour/misinformation topics from more general topics. Finally, the paper makes several suggestions for dealing with the multilingual and code-switched nature of South African Twitter, as well as for the exploration and development of new dynamic topic modeling approaches that could be especially valuable for tracing the development of specific misinformation or rumour narratives over time. The paper presents novel insights and results on the application of a combination of data mining, machine learning and optimisation for addressing the pressing issue of misleading information on social media.

Businesses are increasingly investigating the use of data science and machine learning techniques for market research. This paper investigates the use of topic modeling as a tool for social media market research, specifically the influence and impact of this technology within market research practice. As an example of the use of topic modeling, three different topic modeling algorithms are applied to a single dataset extracted from Reddit, and their performance compared. The latent Dirichlet allocation (LDA) algorithm was trained as a baseline and compared to the Correlated topic model (CTM) and the Gibbs sampling for Dirichlet multinomial mixtures (GSDMM) model. The CTM outperformed the LDA model, while the GSDMM was unable to improve on the baseline. The 25 topics produced by the final CTM were investigated in greater detail and interpreted within the context of market research. Although five of these topics did not prove useful, the remaining topics were easily interpreted and divided into six categories related to (1) features, (2) software, (3) acquisition, (4) workouts, (5) physical design, and (6) physiological monitoring. Each category’s topics were able to provide valuable insight regarding consumers’ opinions about and experiences of the related product.

Machine Learning (ML) algorithms can analyze large amounts of traffic data, learn from patterns and past behaviors, and provide insights into the current and future traffic flow. ML can also optimize traffic management, including traffic signal control, route optimization, and demand forecasting. Traffic prediction is a key application of ML in traffic management, with studies showing that ML outperforms traditional methods in predicting traffic congestion. ML is an effective tool for managing traffic, particularly for projecting traffic demand, predicting traffic congestion, and optimising routes. Studies have revealed that ML is more efficient than conventional techniques in these areas, leading to decreased journey times, improved traffic flow, and better traffic management in general. As the demand for efficient and sustainable transportation systems rises, ML integration in traffic management is expected to be vital in addressing these requirements. Nevertheless, there are obstacles and restrictions that must be overcome, such as shortcomings in the reliability of data and model interpretability. Despite these challenges, ML has the potential to mitigate traffic congestion and enhance urban mobility in smart and sustainable cities. Further research is needed to address these challenges and fully realize the potential of ML in traffic management.

The vagueness of the concept of resilience makes it difficult to define unanimously, and it becomes even more problematic when it comes to measuring it, while urban resilience metrics can be considered as key indicators transmitting vital information to the decision makers on the observed characteristics about the city. The motivations and goals of such a metric are as different as the proponents who defend them. As for cities, due to growing urbanization in a global context of climate change, the concept of urban resilience is essential and requires scientific attention backed by a methodology with an operational aim. Based on 3D spatio-temporal forecasting algorithms, this paper revisits the concept and presents a novel approach to measuring and computing key metrics of resilience applied to urban systems. Some results show that spatio-temporal forecasting algorithms can significantly improve the accuracy and timeliness of urban resilience metrics compared to traditional methods. Our methodology can help urban planners and policymakers make more informed decisions and enhance the resilience of urban systems. However, the methodology also has limitations and challenges, such as data quality issues and algorithmic complexity, that require further research. This paper contributes to the literature on urban resilience and spatio-temporal forecasting by providing a comprehensive framework for measuring and forecasting key metrics of urban resilience using advanced computational methods.

This paper proposes a framework for utilising continuous metaheuristic algorithms for solving drone delivery scheduling problems with interceptions. The use of the framework is illustrated through the application of a particle swarm optimisation-based algorithm, a differential evolution-based algorithm, and a covariance matrix adaptation-evolution strategy for Moremi’s travelling salesman problem with drone with interceptions (TSPDi) [1]. A single drone can be launched to make a delivery while a truck is delivering packages to customers. The drone then either intercepts the purpose-built truck after completing a delivery when possible or meets up with the truck at the next customer location. The optimisation algorithms are tested on benchmark datasets and compared against Moremi’s ant colony optimisation-based drone scheduling algorithm. Algorithm performance is measured with regards to total delivery time and total truck travel distance. An in-depth analysis of algorithm performance is conducted in the form of an efficiency study. The efficiency study specifically considers the number of duplicate and feasible solutions as the different algorithms handle the TSPDi constraints. It is shown that the metaheuristics are especially useful when a smaller number of deliveries are required, whereas performance deteriorates significantly when the number of nodes to be visited grows.The contributions of this paper include the only in-depth analysis of the performance of continuous-based metaheuristics in drone scheduling literature, as well as the first framework for utilising continuous-based metaheuristics for solving drone delivery scheduling problems with interceptions. This paper contributes to the computational sciences by using advanced computing capabilities for decision support to improve the performance of an intelligent transport system.

The paper presents the results of a pilot study of the possibility of using a value-based approach to managing city development programs. The value model of a person includes a static description of the values, motivations and expectations of a person, taking into account his social, demographic group and life situation. The value model of life in the city involves the description of transitions between motivated and non-motivated social groups. City development management is considered from the point of view of the composition of development projects, which requires the formation of a way of presenting the project in the “language” of a value management model, which includes an assessment of the timeliness of the city development project. The paper proposes to determine the values of residents through social media profiles and comments generated by users. Values are described using the Schwartz method. The work contains comments from sources - Google Maps, Telegram, Vkontakte. To validate the information received, a survey of residents was conducted, to compare a real person with a virtual profile, using the revealed correlations, it was concluded that with the help of information from social networks, it is possible to determine values and use them to manage city development programs.

ODEs are a useful mathematical tool for modeling dynamic systems in different fields, such as physics, engineering, biology, and economics. They can provide insights into the behavior of complex systems over time. However, creating ODE models can be difficult and requires expertise in the subject matter and mathematical techniques. This paper presents the OdeShell, a command line interface that enables users to build and simulate ODE models while examining their behavior under diverse circumstances. OdeShell is a valuable addition to the ODE modeling domain, with the capability to ease the development of intricate models in various fields. The tool accommodates novice and proficient modelers, giving them a flexible and user-friendly environment to build and test ODE models. We elaborate on the principal features and functionality of OdeShell and demonstrate its utility in developing ODE models through small examples. Additionally, we discuss the ODE language, emphasizing its syntax and meaning.

In this paper, we propose an algorithm for multiple integer divisions with an invariant dividend and monotonically increasing or decreasing divisors. In such multiple integer divisions, we show that if the dividend and divisors satisfy a certain condition, then if only one quotient is calculated by division first, the remaining quotients can be obtained by correcting the previously calculated quotients at most once. The proposed algorithm is up to approximately 1.90 and 1.85 times faster than the 64-bit unsigned integer division instruction of the Intel 64 architecture and Intel Short Vector Math Library (SVML) on the Intel Xeon Platinum 8368 processor, respectively.

Current galaxy classification studies are usually conducted on small, expert-classified datasets, constrained within a low redshift (z) range. Lower redshift implies better image quality – the lower the z value, the closer the object is, thus more features and details can be observed. Additionally, various augmentation methods are used to further improve classification accuracy, however, there is a lack of studies measuring their impact on other metrics. Therefore, we study the impact of augmentation techniques using the largest dataset that covers a broad redshift range (315,942 galaxies, $$0 < z \leqslant 0.28$$ 0 < z ⩽ 0.28 ). We provide comparable evidence that for binary galaxy image classification, common image augmentation techniques – rotation, zoom and flipping – increase accuracy and F1 score, unless the model is underfitting. The most significant increase was observed on the ResNet50 model, for which the accuracy increased from 93.53% to 95.21%, and the F1 score – from 88.66% to 91.82%. Additionally, combining the aforementioned techniques with random noise stabilises a model by significantly decreasing the spread of metrics – for ResNet50, the standard deviation of the F1 score decreased by more than 9 times.

Recently, as the application range of high-end Organic Light Emitting Diodes (OLED) displays has expanded, the pixel area has become larger and panel architecture complexity has been raised to increase the lifespan and efficiency of the panel. At the same time, the remaining moisture in the organic material of the panel increases, and pixel shrinkage defects in which the pixel edge emission area is reduced by cathode oxidation due to moisture diffusion within the panel are increasing. Therefore time and cost are incurred in the process of design and manufacturing process enhancing to improve reliability of the display panel. In this study, we introduce an analysis model that can quantify pixel shrinkage and predict the occurrence of defects in advance. In order to make up for the shortcomings of existing Finite Element Method (FEM) and numerical analysis models, product design specifications were applied through layout-based 3D geometry, and the degree of curing according to the heat treatment conditions of organic materials was calculated and the physics of moisture diffusion between subsequent cleaning and oven process were also applied in the model. The degree of moisture absorption during the waiting time between processes, one of the main factors affecting defects was applied in connection with the curing rate of organic materials. In addition, the residual concentration after moisture diffusion in the final deposition process was quantified and matched with the actual shrinkage defect occurrence level to optimize design and process factors of the model. As a result of verification based on 20 evaluation models, the consistency of about 0.96 based on R2 was confirmed.

Conversion RGB to wavelength is not a simple problem. This contribution describes a simple method for wavelength extraction for colors given by the RGB triplet. The method is simple and accurate, based on known RGB values of the rainbow. It also respects different saturation of a color.

Proper computer-assisted detection (CADe) based on machine learning (ML) is a hot research topic in healthcare. Thousands of studies are published yearly focusing on enhancing the performance of ML-based models, but few tackle the challenge of deploying them efficiently. This paper focuses on designing an effective graphical user interface (GUI) tool, AI Radiologist, that clinicians can use during pre-operative planning to segment different liver tissues (parenchyma, tumors, and vessels). The tool employs convolutional neural networks (ConvNets) for liver tissue segmentation. This helps increase the success rate of any operation through meticulous pre-planning that allows surgeons to prepare well and plan for worst-case scenarios that might occur during surgery. AI Radiologist, an offline system application, utilizes three ConvNet models trained to segment all liver tissues. We use the PyQt5 Python module for the GUI to create a single-page application. The output of the AI Radiologist application is the liver, tumors, and vessels 2D slices and the 3D interpolation in .obj and .mtl format. The 3D interpolation can be visualized as a 3D liver object on any 3D-friendly software or 3D printed. Creating the AI Radiologist provides clinicians with a user-friendly GUI tool for liver tissues’ segmentation and 3D interpolation, employing state-of-the-art models for all tissues’ segmentation processes. Clinicians can select the volume(s) and the pre-trained models, and the AI Radiologist will take care of the rest.

The paper examines the content and structure of professional communicative practices of helping specialists in digital professional communities, and also identifies psychological influence techniques used to manage interaction with the audience. A platform with profiles of psychologists offering their services was selected for the study. With the help of parsing methods, all the specialists' questionnaires were saved. Further, thematic modeling was carried out using LDA methods. 10 main topics and 55 unique terms describing the content of the professional discourse of helping specialists were identified. As a result, the components of the communicative strategy of using digital resources were evaluated: the presence of professional vocabulary, basic terminology, descriptions of transformational techniques, the presence of a product image, an image of the result, a description of the target audience, the presence of suggestive manipulative technologies. The model of monitoring the professional digital community as a whole, the possibilities of the platform for implementing elements of the communicative strategy of specialists were also described..

Modeling socioeconomic dynamics has always been an area of focus for urban scientists and policymakers, who aim to better understand and predict the well-being of local neighborhoods. Such models can inform decision-makers early on about expected neighborhood performance under normal conditions, as well as in response to considered interventions before official statistical data is collected. While features such as population and job density, employment characteristics, and other neighborhood variables have been studied and evaluated extensively, research on using the underlying networks of human interactions and urban structures is less common in modeling techniques. We propose using the structure of the local urban mobility network (weighted by commute flows among a city’s geographical units) as a signature of the neighborhood and as a source of features to model its socioeconomic quantities. The network structure is quantified through node embedding generated using a graph neural network representation learning model. In the proof-of-concept task of modeling the location’s median income and housing profile in two different cities, such network structure features provide a noticeable performance advantage compared to using only the other available social features. This work can thus inform researchers and stakeholders about the utility of mobility network structure in a complex urban system for modeling various quantities of interest.

Supply chains are very complex systems and their correct and efficient management represents a fundamental challenge, in which the practical needs of the corporate world can find answers together with the advanced skills of the academic world. This paper fits exactly in this area. In particular, starting from a project by the company Code Architects, we will illustrate how it is possible to make forecasts on shipments with machine learning tools, which can support business decisions.

The dialect of a machine learning model is comprised of the features encountered during training. Nonetheless, as time passes, a deployed machine learning model may encounter certain features for the first time. In conventional machine learning approaches, newly observed features are typically discarded during testing data sample preprocessing. In lifelong machine learning, newly observed features may have appeared in the feature space of previously learned tasks; consequently, the knowledge associated with those features present in the knowledge base is incorporated to handle these features. However, there may be some features that have yet to appear in the knowledge base; lifelong machine learning also discards such features. Features that were not seen before are called unseen features. In this paper, we propose an enhanced lifelong machine learning framework for handling unseen features during the testing phase that incorporates relative knowledge. To extract relative knowledge, we retrieve semantically similar features using a language model. In addition, semantically similar features are examined in the knowledge base, and the knowledge of those present in the knowledge base is incorporated in order to deal with unseen features. Experiments conducted on the Amazon review dataset indicate that the proposed method outperforms three baselines and is competitive with state-of-the-art methods.

The emergence of information and mobile technology has had a profound impact on modern life, altering the manner in which individuals communicate, access information, entertain themselves, and conduct business. This development has created new opportunities for researchers to access datasets that were not accessible to prior generations of scholars. Historically, studies of urban commuting have relied on census data, which portrays commute patterns as a static number that are updated every few years. However, over the last two decades, the advent of mobile phone datasets has facilitated new research avenues. This study employs mobile phone mobility data to define a signature of urban districts within the Czech city of Brno, utilizing it for the purpose of urban zoning. The proposed signatures provide a tangible classification of neighborhoods with potential applications in urban and transportation planning. This approach is demonstrated using mobile data of 13 thousand inhabitants of Brno and can be applied to other cities wherever similar data is available.

The scientific objective of the study is to identify the qualification prerequisites of local government management as an element of the economic and social pillar of socially responsible action, based on the analysis of theoretical approaches to socially responsible action in the public sector and secondary research in the environment of local governments. The aim was to identify the real impact of education on selected indicators of economic performance of local government - an element of the economic pillar of social responsibility. The subject of the scientific study is the education (theoretical as well as practical component) of the statutory as well as local government councillors and their impact on selected economic indicators of local government performance such as long-term and short-term liabilities, capital expenditures, non-current assets, liquidity, current account balance, investment intensity, the size of the municipality’s net assets per capita, economic performance as well as the overall financial health of the local government. The subject was also an examination of the impact of education, as an element of the social pillar of social responsibility, on another, economic pillar of social responsibility - efficient and responsible management.

The problem of efficient and socially responsible management of small municipalities is a discussed topic in many countries of the world. Applying the concept of social responsibility in the management of self-government strengthens the image, increases transparency and contributes to the interest of citizens in governance in self-government. The aim of the paper is to analyze the problems of small municipalities in Slovakia and abroad, and to propose partial solutions mainly in the economic and administrative area based on the principles of social responsibility, inter-municipal cooperation and possible positive impacts on management due to economies of scale. As small municipalities are unable to take advantage of the effects of economies of scale, certain activities (mainly municipal administration) are more expensive than larger ones. However, before the idea of ​​merging municipalities is approached, it is first necessary to apply tools that will enable municipalities to perform certain activities cheaper, faster, or of better quality. The essence of these tools must be based on mutual cooperation, specialization, joint use of human, material, and financial resources and subsequent use of the effects that such cooperation brings (including marketing - improving the image of the community, improving communication channels, benchmarking, etc.).