With the ease of access to social media platforms, the spread of fake news has become a growing concern in today’s society. Classifying fake news is an important task, as it can help prevent its negative impact on individuals and society. In this work, an end-to-end framework for fake news detection is developed by utilizing the power of adversarial training to make the model more robust. This framework is named “ANN: Adversarial News Net”. The performance of ANN is evaluated using four publicly available datasets, and it is found to outperform previous studies after adversarial training. Furthermore, emoticons have been extracted from the dataset to understand their meanings in relation to fake news. This information is then fed into the model, which helped to improve its performance in classifying fake news. The proposed framework has the potential to be used as a tool for detecting fake news in real-time, thereby mitigating its harmful effects on society.

The purpose of text classification is to label the text with known labels. In recent years, the method based on graph neural network (GNN) has achieved good results. However, the existing methods based on GNN only regard the text as the set of co-occurring words, without considering the position information of each word in the statement. Meanwhile, this method mainly extracts node features, but neglects the use of edge features between nodes. To solve these problems, a new text classification method, graph convolutional network using positions and edges (PEGCN), is proposed. In the word embedding section, a positional encoding input representation is employed to enable the neural network to learn the relative positional information among words. Meanwhile, the dimension of the adjacency matrix is increased to extract the multi-dimensional edge features. Through experiments on multiple text classification datasets, the proposed method is shown to be superior to the traditional text classification method, and has achieved a maximum improvement of more than 4%.

The analysis of tourist accommodation bookings provides valid information for the management of these establishments. The objective of this work is to analyze the performance of different Machine Learning techniques for the prediction of booking cancellations, as well as to analyze possible patterns in the study data. For this purpose, the following supervised learning methods are used: Multilayer Perceptron Neural Network, Radial Basis Function Neural Network, Decision Tree, Random Forest, AdaBoost and XgBoost, analyzing the performance of these techniques. The dataset used corresponds to the bookings of a resort hotel and a city hotel located in Portugal. As a result, the study compares the classification methods applied and identifies those with better performance, proving that Machine Learning techniques generate reliable forecasts for the management of the tourism industry.

This paper proposes a novel algorithm that accurately predicts market trends and trading entry points for US 30-year-treasury bonds using a hybrid approach of 1-Dimensional Convolutional Neural Network (1DCNN), Long-Short Term Memory (LSTM), and XGBoost algorithms. We compared the performance of various strategies using 1DCNN and LSTM and found that existing state-of-the-art methods based on LSTM have excellent results in market movement prediction tasks, but the effectiveness of 1DCNN and LSTM in terms of trading entry point and market perturbations has not been studied thoroughly. We demonstrate, through experiments that our proposed 1DCNN-BiLSTM-XGBoost algorithm combined with moving averages crossover effectively mitigates noise and market perturbations, leading to high accuracy in spotting trading entry points and trend signals for US 30-year-treasury-bonds. Our experimental study shows that the proposed approach achieves an average of 0.0001% Root Mean Squared Error and 100% R-Square, making it a promising method for predicting the market trends and trading entry points.

The paper proposes an alternative approach to improve the performance of image retrieval. In this work, a framework for image retrieval based on machine learning and semantic retrieval is proposed. In the preprocessing phase, the image is segmented objects by using Graph-cut, and the feature vectors of objects presented in the image and their visual relationships are extracted using R-CNN. The feature vectors, visual relationships, and their symbolic labels are stored in KD-Tree data structures which can be used to predict the label of objects and visual relationships later. To facilitate semantic query, the images use the RDF data model and create an ontology for the symbolic labels annotated. For each query image, after extracting their feature vectors, the KD-Tree is used to classify the objects and predict their relationship. After that, a SPARQL query is built to extract a set of similar images. The SPARQL query consists of triple statements describing the objects and their relationship which were previously predicted. The evaluation of the framework with the MS-COCO dataset and Flickr showed that the precision achieved scores of 0.9218 and 0.9370 respectively.

This article presents the architecture, design and validation of a microservice orchestration approach, that improves the flexibility of heterogeneous microservice-based platforms. Improving user experience and interaction, for time-critical applications are aspects that were primary objectives for the design of the architecture. Each microservice can provide its own embedded user interface component, also decentralizing it and, in consequence, improving the loosely coupled approach to the architecture. Obtained results are promising, with high throughput and low response times. Also, a key finding was the introduction of benchmarking as a new step in the development lifecycle of performance-critical software components, with an example of how it can be applied within an Agile methodology. Further research is proposed to improve the results and raise the final technology readiness level of the system. Obtained results already make the approach a candidate and viable alternative to classical service composers.

The Siamese tracker consists of two components: a classification and a regression networks. Despite their different roles, most Siamese trackers have similar feature fusion modules in the two networks, leading to the neglect of their unique characteristics. In this work, we experimentally discover that the two networks place different levels of emphasis on different types of information. Specifically, regression tends to rely on semantic information, while classification places more emphasis on global information. Therefore, we propose a new tracking structure named SGTrack, which includes a semantic augmentation fusion (SAF) for regression and a global relevance fusion (GRF) for classification. It allows us to unlock the full potential of both networks. The experimental results of our method on five benchmarks provide evidence of a notable improvement in tracking performance, while preserving real-time speed.

The repair process of devices is an important part of the business of many original equipment manufacturers. The consumption of spare parts, during the repair process, is driven by the defects found during inspection of the devices, and these parts are a big part of the costs in the repair process. In previous work we proposed a data-driven method for Supply Chain Control Tower solutions to provide support for the automatic check of spare parts consumption in the repair process. In this paper, we continue our investigation of a multi-label classification problem and explore alternatives in the learning-to-rank approach, where we simulate the passage of time using more data while training and comparing hundreds of Machine Learning models to provide an automatic check in the consumption of spare parts. We investigate the effects of different train set sizes, retraining intervals, models and hyper-parameter search using Bayesian Optimization. The results show that we were able to improve the trained models and achieve a higher mean NDCG@20 score of 86% when ranking the expected parts. Focusing on the most recent data, we achieve a NDCG@20 score of 90%, while obtaining a ratio of marked parts of just 4% of the consumed parts for use in alert generation.

Developing analytical systems imposes several challenges related not only to the amount and heterogeneity of the involved data but also to the constant need to readapt and evolve to overcome new business challenges. Data is a determinant factor in the success of analytical and decision-making applications, being its nature, availability, and quality, crucial aspects for planning and structuring populating analytical systems. Today’s users are more demanding, requiring adaptable and flexible analytical applications, which impose serious challenges on ETL systems design and development for ensuring flexible and robust data populating services, operating 24/7, and managing and processing large volumes of data. Thus, we should design and implement ETL processes using innovative and up-to-date approaches, having real application evidence. In this paper, we present a service-oriented implementation for ETL design and development. We mapped and implemented some of the most conventional ETL processes in a service-oriented architecture, for demonstrating the application and benefits that this kind of approach will provide to ETL systems project development.

The growing use of technology and social media has resulted in the emergence of digital influencers, a new profession capable of changing the mentalities and behaviours of those who follow them. This study arises to better understand the potential impact digital influencers might have on the Portuguese population’s purchase behaviour and patterns, and for this purpose, seven hypotheses were formulated. An online questionnaire was conducted to respond to these theoretical assumptions and collected data from 175 respondents. A total of 129 valid answers were considered. It was possible to conclude that purchase intention does not necessarily translate into a purchase action. It was also concluded that the relationship between social network use and the purchase of products/services recommended by influencers is only significant for Instagram. Furthermore, individuals’ Generation is not significantly linked with purchasing a product/service recommended by influencers. Furthermore, a small percentage of respondents have also identified themselves as impulsive shoppers and perceived Instagram as their favourite social network. With the results of this study, it is also possible to state that the influencer’s opinion was classified as the last factor considered in the purchase decision process. Additionally, there is a weak negative association between purchasing a product/service recommended by influencers with sponsorship disclosure and remunerated partnership, which decreases credibility and discourages purchasing.

India sees a 23% rise in podcast listening after the Covid-19 pandemic. The pandemic and screen fatigue led people to seek out old favourite audio podcasts. Podcast genre classification allows listeners to compile a playlist of their favourite tracks; this also helps podcast streaming services provide recommendations to users based on the genre of the podcasts they enjoy. After the COVID-19 pandemic, the need for educational content in all forms, including podcasts, has skyrocketed, making it even more crucial to anticipate the genre of educational podcasts. Educational podcasts are a sub-genre of the broader education genre and typically involve audio recordings of discussions, lectures, or interviews on educational topics. Education podcast genre prediction is required to efficiently classify and arrange educational content and make it simpler for listeners to access and absorb pertinent information. This study focuses on Podcast Genre Prediction, specifically for the Hindi language. In our study, our developed PodGen dataset is used, which consists of 550 podcasts of 5 minutes each and have a total of 26,867 sentences, where every podcast has been manually annotated into one of the four genre categories (Horror, Motivational, Crime, and Romance). The performance comparison of state-of-the-art machine learning techniques on the PodGen dataset is used to demonstrate accuracy. The best performance was observed in the case of the Support Vector Classifier model with balanced accuracy:82.42%, precision (weighted):83.09%, recall (weighted):82.42%, and F1 score(weighted):82.39% on testing data.

Analysis of human emotions from multimodal data for making critical decisions is an emerging area of research. The evolution of deep learning algorithms has improved the potential for extracting value from multimodal data. However, these algorithms do not often explain how certain outputs from the data are produced. This study focuses on the risks of using black-box deep learning models for critical tasks, such as emotion recognition, and describes how human understandable interpretations of these models are extremely important. This study utilizes one of the largest multimodal datasets available - CMU-MOSEI. Many researchers have used the pre-extracted features provided by the CMU Multimodal SDK with black-box deep learning models making it difficult to interpret the contribution of individual features. This study describes the implications of individual features from various modalities (audio, video, text) in Context-Aware Multimodal Emotion Recognition. It describes the process of curating reduced feature models by using the GradientSHAP XAI method. These reduced models with highly contributing features achieve comparable and even better results compared to their corresponding all feature models as well as the baseline model GraphMFN proving that carefully selecting significant features can help improve the model robustness and performance and in turn make it trustworthy.

:30.0 The development of machine learning technologies are broadly changing how humans interact with their environments across all sectors. In industrial settings, this is referred to as the fourth industrial revolution, Industry 4.0, and encompasses several technologies that are pushing the boundaries of industrial automation. In this study, a general industrial process optimization (GIPO) methodology is formulated in the context of Industry 4.0 and tested on an industrial Injection Molding Machine (IMM). GIPO aims to encourage the practical inclusion of industrial artificial intelligence at all levels of the manufacturing process while enabling industrial equipment to adapt to a changing processing environment. Special attention is given to the generality of the methodology so that it can be extended to other applications. In the example case study presented here, GIPO combines nearest neighbors classification and nearest neighbors optimization methods to effectively optimize an Injection molding process. Practical implementation conducted on the IMM demonstrates a novel methodology to leverage data mining and machine learning methods in a real-world setting to improve the overall performance regarding production time, energy cost, and production quality.

The paper deals with the development of 2-link lower limb extremity manipulator for rehabilitation purposes in bio-medical engineering. The musculoskeletal modeling was performed on the collected human gait data to obtain the joint angles and driving force values for hip and knee movement. The results obtained from the proposed controller were verified using Matlab-Simulink. The robust adaptive RBF neural network and fuzzy sliding mode controller were developed to track the desired trajectory. The OpenSim software was utilized to perform inverse kinematics and inverse dynamics on the musculoskeletal model which is scaled according to the physical dimensions of the experimental subject. The simulation results stated that SMC-RBF network has better tracking rate, less chattering problems and handle unknown uncertainties well. Finally, the results obtained were compared with the existing method with the help of performance index parameters to prove their effectiveness.

Competitive Intelligence allows an organization to keep up with market trends and foresee business opportunities. This practice is mainly performed by analysts scanning for any piece of valuable information in a myriad of dispersed and unstructured sources. Here we present MapIntel, a system for acquiring intelligence from vast collections of text data by representing each document as a multidimensional vector that captures its own semantics. The system is designed to handle complex Natural Language queries and visual exploration of the corpus, potentially aiding overburdened analysts in finding meaningful insights to help decision-making. The system searching module uses a retriever and re-ranker engine that first finds the closest neighbors to the query embedding and then sifts the results through a cross-encoder model that identifies the most relevant documents. The browsing or visualization module also leverages the embeddings by projecting them onto two dimensions while preserving the multidimensional landscape, resulting in a map where semantically related documents form topical clusters which we capture using topic modeling. This map aims at promoting a fast overview of the corpus while allowing a more detailed exploration and interactive information encountering process. We evaluate the system and its components on the 20 newsgroups dataset, using the semantic document labels provided, and demonstrate the superiority of Transformer-based components. Finally, we present a prototype of the system in Python and show how some of its features can be used to acquire intelligence from a news article corpus we collected during a period of 8 months.

Psychological stress has evolved as an important health concern across the globe. The vulnerability to stress and the ramifications of it have only worsened during the time of the COVID-19 pandemic. This necessitates a timely diagnosis of stress before the condition progresses to chronicity. In this context, the popularity of social media like Twitter, where large numbers of users share opinions without any social stigma, has emerged as a major resource of human opinions. This has led to an increased research interest in social media-based stress detection techniques. However, tweet-level stress detection techniques in the literature have left a void in leveraging the text information in tweets, especially the presence of sarcastic expressions in the tweet’s text content. To this end, a novel method called “Sarcasm-based Tweet-Level Stress Detection” (STSD) is proposed in this work with the modification of the logistic loss function to detect tweet-level stress by availing the information of sarcasm that exists in the tweet-content. The principle of the STSD model is to minimise the loss for non-sarcastic tweets while maximising the loss for sarcastic tweets. Furthermore, an extensive preprocessing and dimensionality reduction is performed using kernel principal component analysis (kernel PCA) to improve the performance by reducing the dimensions. The experimental results show that the proposed STSD model, when applied along with kernel PCA, records a significant improvement in accuracy by a minimum of 5.25% and a maximum of 9.19% over baseline models. Also, there is an increment in F1-score by at least 0.085 points and a maximum of 0.164 points when compared to the baseline models.