Progress in Artificial Intelligence

This paper presents the first results on the validation of a new Adaptive E-learning System, focused on providing personalized learning to secondary school students in the field of education about AI by means of an adaptive interface based on a 3D robotic simulator. The prototype tool presented here has been tested at schools in USA, Spain, and Portugal, obtaining very valuable insights regarding the high engagement level of students in programming tasks when dealing with the simulated interface. In addition, it has been shown the system reliability in terms of adjusting the students’ learning paths according to their skills and competences in an autonomous fashion.

This paper presents the design and evaluation of Gamified CollectiveEyes that is a digital infrastructure to collectively share human eyes. Gamified CollectiveEyes collects people’s viewpoints in the world anywhere at all times, and a user sees several collected viewpoints simultaneously in a 3D virtual space. For navigating human viewpoints collected by Gamified CollectiveEyes, we propose a novel abstraction named topic channel in the paper, where a user can choose appropriate viewpoints and hearings that he/she wants to see. After presenting an overview of Gamified CollectiveEyes, we show two user studies to investigate potential opportunities and pitfalls of Gamified CollectiveEyes: the first user study is to investigate the human motivation mechanism to offer their viewpoints and the second user study is to investigate the configuration to present multiple viewpoints. We also show the limitation and future work of the current development of Gamified CollectiveEys.

Poor and sedentary lifestyles combined with bad dietary habits have an impact on our health. Nowadays, diet-related diseases have become a major public health issue, threatening the sustainability of healthcare systems, and new strategies to promote better food intake are now being explored. In this context, the use of ontologies has gained importance over the past decade and become more prevalent. By incorporating ontologies in the healthcare domain, artificial intelligence (AI) can be enhanced to better support healthcare systems dealing with chronic diseases, such as obesity and diabetes requiring long-term progress and frequent monitoring. This is especially challenging with current resource inefficiency; however, recent research suggests that incorporating ontology into AI-based technological solutions can improve their accuracy and capabilities. Additionally, recommendation and expert systems benefit from incorporating ontologies for a better knowledge representation and processing to increase success rates. This study outlines the development of an ontology in the context of food intake to manage and monitor patients with obesity, diabetes, and those using tube feeding. A standardized vocabulary for describing food and nutritional information was specified to enable the integration with different healthcare systems and provide personalized dietary recommendations to each user.

We are used to seeing the manifestations of various emotions in humans, but animals also show emotions. A better understanding of animal emotions is closely related to creating animal welfare. Research in this direction may impact other ways to improve the lives of domestic and farm animals or animals in captivity. In addition, better recognition of negative emotions in animals can help prevent unwanted behaviour and health problems caused by long-term increased levels of stress or other negative emotional states. Research projects focused on the emotional needs of animals can benefit animals and contribute to a more ethical and sustainable relationship between humans and animals.This article is focused on the one hand on the description of the system that was created in the previous related research for monitoring the vital functions of animals, and on the other hand, especially on the investigation of the possibilities of how the given system can be used to identify the emotional states of animals.

As the frontier of machine learning applications moves further into human interaction, multiple concerns arise regarding automated decision-making. Two of the most critical issues are fairness and data privacy. On the one hand, one must guarantee that automated decisions are not biased against certain groups, especially those unprotected or marginalized. On the other hand, one must ensure that the use of personal information fully abides by privacy regulations and that user identities are kept safe. The balance between privacy, fairness, and predictive performance is complex. However, despite their potential societal impact, we still demonstrate a poor understanding of the dynamics between these optimization vectors. In this paper, we study this three-way tension and how the optimization of each vector impacts others, aiming to inform the future development of safe applications. In light of claims that predictive performance and fairness can be jointly optimized, we find this is only possible at the expense of data privacy. Overall, experimental results show that one of the vectors will be penalized regardless of which of the three we optimize. Nonetheless, we find promising avenues for future work in joint optimization solutions, where smaller trade-offs are observed between the three vectors.

Competition in Artificial Intelligence (AI) technologies is at its fiercest, pushing companies to move fast and sometimes cut corners regarding the risks to human rights and other societal impacts. Without simple methodologies and widely accepted instruments is hard for an organization to adopt a safe pace on how to develop and deploy AI in a trustworthy way. This paper presents a Maturity Model for Responsible AI, inspired in the EU Ethics Guidelines for Trustworthy AI and other principles and codes of conduct. The core component is a self-assessment tool that generates a roadmap for organizations to improve their approach to AI related development, enabling a positive effect in their business value. It includes requirements to achieve Trustworthy AI, and the methods and key practices that will enable the principles outlined. The result is a consistent and horizontal approach to all industries and functions, taking into consideration that a simple and generic Maturity Model, specific for Responsible AI, is still not available. The model presented in this paper is purposed to fill that gap. The model was pre-tested in two organizations, improved and pre-tested again. Results are presented in this paper. The next phase is to apply the model in several other organizations, and conduct interviews at different hierarchical levels and promote case study discussions. Its final version is planned to be published at the end of 2023.

ML/AI is the field of computer science and computer engineering that arguably received the most attention and funding over the last decade. Data is the key element of ML/AI, so it is becoming increasingly important to ensure that users are fully aware of the quality of the datasets that they use, and of the process generating them, so that possible negative impacts on downstream effects can be tracked, analysed, and, where possible, mitigated. One of the tools that can be useful in this perspective is dataset documentation. The aim of this work is to investigate the state of dataset documentation practices, measuring the completeness of the documentation of several popular datasets in ML/AI repositories. We created a dataset documentation schema-the Documentation Test Sheet (dts)-that identifies the information that should always be attached to a dataset (to ensure proper dataset choice and informed use), according to relevant studies in the literature. We verified 100 popular datasets from four different repositories with the dts to investigate which information were present. Overall, we observed a lack of relevant documentation, especially about the context of data collection and data processing, highlighting a paucity of transparency.

Artificial intelligence (AI) has been widely used in many fields, from intelligent virtual assistants to medical diagnosis. However, there is no consensus on how to deal with ethical issues. Using a systematic literature review and an analysis of recent real-world news about AI-infused systems, we cluster existing and emerging AI ethics and responsibility issues into six groups - broken systems, hallucinations, intellectual property rights violations, privacy and regulation violations, enabling malicious actors and harmful actions, environmental and socioeconomic harms - discuss implications, and conclude that the problem needs to be reflected upon and addressed across five partially overlapping dimensions: Research, Education, Development, Operation, and Business Model. This reflection may be relevant to caution of potential dangers and frame further research at a time when products and services based on AI exhibit explosive growth. Moreover, exploring effective ways to involve users and civil society in discussions on the impact and role of AI systems could help increase trust and understanding of these technologies.

The financial industry is a major player in the digital landscape and a key driver of digital transformation in the economy. In recent times, the financial sector has come under scrutiny due to emerging financial crises, particularly in high-risk areas like credit scoring models where standard AI models may not be fully reliable. This highlights the need for greater accountability and transparency in the use of digital technologies in Fintech. In this paper, we propose a novel approach to enhance the interpretability of AI models by knowledge augmentation using distillation methods. Our aim is to transfer the knowledge from black-box models to more transparent and interpretable models, e.g., decision-trees, enabling a deeper understanding of decision patterns. We apply our method to a credit score problem and demonstrate that it is feasible to use white-box techniques to gain insight into the decision patterns of black-box models. Our results show the potential for improving interpretability and transparency in AI decision-making processes in Fintech scenarios.

Attention-based agents have had much success in many areas of Artificial Intelligence, such as Deep Reinforcement Learning. This work revisits two such architectures, namely, Deep Attention Recurrent Q-Networks (DARQNs) and Soft Top-Down Spatial Attention (STDA) and explores the similarities between them. More specifically, this work tries to improve the performance of the DARQN architecture by leveraging elements proposed by the STDA architecture, such as the formulation of its attention function which also includes the incorporation of a spatial basis into its computation. The implementation tested, denoted Deep Attention Recurrent Actor-Critic (DARAC), uses the A2C learning algorithm. The results obtained seem to suggest that the performance of DARAC can be improved by the incorporation of some of the techniques proposed in STDA. Overall, DARAC showed competitive results when compared to STDA and slightly better in some of the experiments performed. The Atari 2600 videogame benchmark was the testbed used to perform and validate all the experiments.

This work explores pre-training as a strategy to allow reinforcement learning (RL) algorithms to efficiently adapt to new (albeit similar) tasks. We argue for introducing variability during the pre-training phase, in the form of augmentations to the observations of the agent, to improve the sample efficiency of the fine-tuning stage. We categorize such variability in the form of perceptual, dynamic and semantic augmentations, which can be easily employed in standard pre-training methods. We perform extensive evaluations of our proposed augmentation scheme in model-based algorithms, across multiple scenarios of increasing complexity. The results consistently show that our augmentation scheme significantly improves the efficiency of the fine-tuning to novel tasks, outperforming other state-of-the-art pre-training approaches.

Neural networks have achieved remarkable success in various applications such as image classification, speech recognition, and natural language processing. However, the growing size of neural networks poses significant challenges in terms of memory usage, computational cost, and deployment on resource-constrained devices. Pruning is a popular technique to reduce the complexity of neural networks by removing unnecessary connections, neurons, or filters. In this paper, we present novel pruning algorithms that can reduce the number of parameters in neural networks by up to 98% without sacrificing accuracy. This is done by scaling the pruning rate of the models to the size of the model and scheduling the pruning to execute throughout the training of the model. Code related to this work is openly available.

This study examines how perturbations in image, audio, and text inputs affect the performance of different classification models. Various perturbators were applied to three seed datasets at different intensities to produce noisy test data. Then, the models’ performance was evaluated on the generated test data. Our findings indicate that there is a consistent relationship between larger perturbations and lower model performance across perturbators, models, and domains. However, this relationship varies depending on the characteristics of the specific model, dataset, and perturbator.

Identifying small dolphin species based on their vocalizations remains a challenging task due to their similar vocal signatures and frequency modulation patterns, particularly when the available data sets are relatively limited. To address this issue, a new feature set has been introduced that focuses on capturing both the predominant frequency range of the vocalizations and other higher level details in the spectral contour, which are valuable for distinguishing between small dolphin species. These features are computed from two distinct representations of the vocalizations: the short time Fourier transform and Mel frequency cepstral coefficients. By utilizing these features with two popular classifiers (K-Nearest Neighbors and Support Vector Machines), a model accuracy of $$95.47\%$$ 95.47 % has been achieved, representing an improvement over previous studies.

Probabilistic supervised learning within the Bayesian paradigm typically use Gaussian Processes (GPs) to model the sought function, and provide a means for securing reliable uncertainties in said functional learning, while offering interpretability. Prediction of the output of such a learnt function is closed-form in this approach. In this work, we present GP based learning of the functional relation between two variables, using various kinds of kernels that are called in to parametrise the covariance function of the invoked GP. However, such covariance kernels are typically parametric in the literature, with hyperparameters that are learnt from the data. Here, we discuss a new nonparametric covariance kernel, and compare its performance against existing non-stationary and stationary kernels, as well as against Deep Neural Networks. We present results on both univariate and multivariate data, to demonstrate the range of applicability of the presented learning scheme.

Quadruped robots are gaining attention in the research community because of their superior mobility and versatility in a wide range of applications. However, they are restricted to procedures that do not need precise object interaction. With the addition of a robotic arm, they can overcome these drawbacks and be used in a new set of tasks. Combining a legged robot’s dextrous movement with a robotic arm’s maneuverability allows the emergence of a highly flexible system, but with the disadvantage of higher complexity of motion planning and control methods. This paper gives an overview of the existing quadruped systems capable of manipulation, with a particular interest in systems with high movement flexibility. The main topics discussed are the motion planning approaches and the selected kinematic configuration. This review concludes that the most followed research path is to add a robotic arm on the quadrupedal base and that the motion planning approach used depends on the desired application. For simple tasks, the arm can be seen as an independent system, which is simpler to implement. For more complex jobs the coupling effects between the arm and quadruped robot must be considered.

Over the past few decades, road transportation emissions have increased. Vehicles are among the most significant sources of pollutants in urban areas. As such, several studies and public policies emerged to address the issue. Estimating greenhouse emissions and air quality over space and time is crucial for human health and mitigating climate change. In this study, we demonstrate that it is feasible to utilize raw GPS data to measure regional pollution levels. By applying feature engineering techniques and using a microscopic emissions model to calculate vehicle-specific power (VSP) and various specific pollutants, we identify areas with higher emission levels attributable to a fleet of taxis in Porto, Portugal. Additionally, we conduct network analysis to uncover correlations between emission levels and the structural characteristics of the transportation network. These findings can potentially identify emission clusters based on the network’s connectivity and contribute to developing an emission inventory for an urban city like Porto.

The accurate estimation of dam natural frequencies and their evolution over time can be very important for dynamic behaviour analysis and structural health monitoring. However, automatic modal parameter estimation from ambient vibration measurements on dams can be challenging, e.g., due to the influence of reservoir level variations, operational effects, or dynamic interaction with appurtenant structures. This paper proposes a novel methodology for improving the automatic identification of natural frequencies of dams using a supervised Convolutional Neural Network (CNN) trained on real preprocessed sensor monitoring data in the form of spectrograms. Our tailored CNN architecture, specifically designed for this task, represents the first of its kind. The case study is the 132 m high Cabril arch dam, in operation since 1954 in Portugal; the dam was instrumented in 2008 with a continuous dynamic monitoring system. Modal analysis has been performed using an automatic modal identification program, based on the Frequency Domain Decomposition (FDD) method. The evolution of the experimental natural frequencies of Cabril dam over time are compared with the frequencies predicted using the parameterized CNN based on different sets of data. The results show the potential of the proposed neural network to complement the implemented modal identification methods and improve automatic frequency identification over time.

In natural phenomena, data distributions often deviate from normality. One can think of cataclysms as a self-explanatory example: rarely occurring events differ considerably from common outcomes. In real-world domains, such tail events are often the most relevant to anticipate, allowing us to take adequate measures to prevent or attenuate their impact on society. However, mapping target values to particular relevance judgements is challenging and existing methods do not consider the impact of bias in reaching such mappings—relevance functions. In this paper, we tackle the issue of uncertainty in non-uniform domain preferences and its impact on imbalanced regression evaluation. Specifically, we develop two methods for assessing the volatility of model performance when dealing with uncertainty regarding the range of target values that are more important to the underlying problem. We demonstrate the importance of our proposed methods in capturing the impact of small changes in relevance assessments of target values and how they may impact experimental conclusions.

Nowadays, all kinds of sensors generate data, and more metrics are being measured. These large quantities of data are stored in large data centers and used to create datasets to train Machine Learning algorithms for most different areas. However, processing that data and training the Machine Learning algorithms require more time, and storing all the data requires more space, creating a Big Data problem. In this paper, we propose simple techniques for reducing large time series datasets into smaller versions without compromising the forecasting capability of the generated model and, simultaneously, reducing the time needed to train the models and the space required to store the reduced sets. We tested the proposed approach in three public and one private dataset containing time series with different characteristics. The results show, for the datasets studied that it is possible to use reduced sets to train the algorithms without affecting the forecasting capability of their models. This approach is more efficient for datasets with higher frequencies and larger seasonalities. With the reduced sets, we obtain decreases in the training time between 40 and 94% and between 46 and 65% for the memory needed to store the reduced sets.

This paper analyses the causal relationship between external events and sports content TV audiences. To accomplish this, we explored external data related to sports TV audience behaviour within a specific time frame and applied a Granger causality analysis to evaluate the effect of external events on both TV clients’ volume and viewing times. Compared to regression studies, Granger causality analysis is essential in this research as it provides a more comprehensive and accurate understanding of the causal relationship between external events and sports TV viewership. The study results demonstrate a significant impact of external events on the TV clients’ volume and viewing times. External events such as the type of tournament, match popularity, interest and home team effect proved to be the most informative about the audiences. The findings of this study can assist TV distributors in making informed decisions about promoting sports broadcasts.

In an automatic music playlist generator, such as an automated online radio channel, how should the system react when a user hits the skip button? Can we use this type of negative feedback to improve the list of songs we will playback for the user next? We propose SkipAwareRec, a next-item recommendation system based on reinforcement learning. SkipAwareRec recommends the best next music categories, considering positive feedback consisting of normal listening behaviour, and negative feedback in the form of song skips. Since SkipAwareRec recommends broad categories, it needs to be coupled with a model able to choose the best individual items. To do this, we propose Hybrid SkipAwareRec. This hybrid model combines the SkipAwareRec with an incremental Matrix Factorisation (MF) algorithm that selects specific songs within the recommended categories. Our experiments with Spotify’s Sequential Skip Prediction Challenge dataset show that Hybrid SkipAwareRec has the potential to improve recommendations by a considerable amount with respect to the skip-agnostic MF algorithm. This strongly suggests that reformulating the next recommendations based on skips improves the quality of automatic playlists. Although in this work we focus on sequential music recommendation, our proposal can be applied to other sequential content recommendation domains, such as health for user engagement.

Machine learning models are widely used in time series forecasting. One way to reduce its computational cost and increase its efficiency is to select only the relevant exogenous features to be fed into the model. With this intention, a study on the feature selection methods: Pearson correlation coefficient, Boruta, Boruta-Shap, IMV-LSTM, and LIME is performed. A new method focused on interpretability, SHAP-LSTM, is proposed, using a deep learning model training process as part of a feature selection algorithm. The methods were compared in 2 different datasets showing comparable results with lesser computational cost when compared with the use of all features. In all datasets, SHAP-LSTM showed competitive results, having comparatively better results on the data with a higher presence of scarce occurring categorical features.

Effective quality control is essential for efficient and successful manufacturing processes in the era of Industry 4.0. Artificial Intelligence solutions are increasingly employed to enhance the accuracy and efficiency of quality control methods. In Computer Numerical Control machining, challenges involve identifying and verifying specific patterns of interest or trends in a time-series dataset. However, this can be a challenge due to the extensive diversity. Therefore, this work aims to develop a methodology capable of verifying the presence of a specific pattern of interest in a given collection of time-series. This study mainly focuses on evaluating One-Class Classification techniques using Linear Frequency Cepstral Coefficients to describe the patterns on the time-series. A real-world dataset produced by turning machines was used, where a time-series with a certain pattern needed to be verified to monitor the wear offset. The initial findings reveal that the classifiers can accurately distinguish between the time-series’ target pattern and the remaining data. Specifically, the One-Class Support Vector Machine achieves a classification accuracy of 95.6 % ± 1.2 and an F1-score of 95.4 % ± 1.3.

In this paper, we explore and compare three recently proposed Automated Machine Learning (AutoML) tools (AutoGluon, H $$_2$$ 2 O, Oracle AutoMLx) to create a single regression model that is capable of predicting smart city energy building consumption values. Using a recently collected one year hourly energy consumption dataset, related with 29 buildings from a Portuguese city, we perform several Machine Learning (ML) computational experiments, assuming two sets of input features (with and without lagged data) and a realistic rolling window evaluation. Furthermore, the obtained results are compared with a univariate Time Series Forecasting (TSF) approach, based on the automated FEDOT tool, which requires generating a predictive model for each building. Overall, competitive results, in terms of both predictive and computational effort performances, were obtained by the input lagged AutoGluon single regression modeling approach.

Several systems that employ machine learning models are subject to strict latency requirements. Fraud detection systems, transportation control systems, network traffic analysis and footwear manufacturing processes are a few examples. These requirements are imposed at inference time, when the model is queried. However, it is not trivial how to adjust model architecture and hyperparameters in order to obtain a good trade-off between predictive ability and inference time. This paper provides a contribution in this direction by presenting a study of how different architectural and hyperparameter choices affect the inference time of a Convolutional Neural Network for network traffic analysis. Our case study focus on a model for traffic correlation attacks to the Tor network, that requires the correlation of a large volume of network flows in a short amount of time. Our findings suggest that hyperparameters related to convolution operations—such as stride, and the number of filters—and the reduction of convolution and max-pooling layers can substantially reduce inference time, often with a relatively small cost in predictive performance.

In recent years, the increasing availability of distributed data has led to a growing interest in transfer learning across multiple nodes. However, local data may not be adequate to learn sufficiently accurate models, and the problem of learning from multiple distributed sources remains a challenge. To address this issue, Machine Learning Data Markets (MLDM) have been proposed as a potential solution. In MLDM, autonomous agents exchange relevant data in a cooperative relationship to improve their models. Previous research has shown that data exchange can lead to better models, but this has only been demonstrated with only two agents. In this paper, we present an extended evaluation of a simple version of the MLDM framework in a collaborative scenario. Our experiments show that data exchange has the potential to improve learning performance, even in a simple version of MLDM. The findings conclude that there exists a direct correlation between the number of agents and the gained performance, while an inverse correlation was observed between the performance and the data batch sizes. The results of this study provide important insights into the effectiveness of MLDM and how it can be used to improve learning performance in distributed systems. By increasing the number of agents, a more efficient system can be achieved, while larger data batch sizes can decrease the global performance of the system. These observations highlight the importance of considering both the number of agents and the data batch sizes when designing distributed learning systems using the MLDM framework.

Learning to forecast spatiotemporal (ST) environmental processes from a sparse set of samples collected autonomously is a difficult task from both a sampling perspective (collecting the best sparse samples) and from a learning perspective (predicting the next timestep). Recent work in spatiotemporal process learning focuses on using deep learning to forecast from dense samples. Moreover, collecting the best set of sparse samples is understudied within robotics. An example of this is robotic sampling for information gathering, such as using UAVs/UGVs for weather monitoring. In this work, we propose a methodology that leverages a neural methodology called Recurrent Neural Processes to learn spatiotemporal environmental dynamics for forecasting from selective samples gathered by a team of robots using a mixture of Gaussian Processes model in an online learning fashion. Thus, we combine two learning paradigms in that we use an active learning approach to adaptively gather informative samples and a supervised learning approach to capture and predict complex spatiotemporal environmental phenomena.

E-commerce has become an essential aspect of modern life, providing consumers worldwide with convenience and accessibility. However, the high volume of short and noisy product descriptions in text streams of massive e-commerce platforms translates into an increased number of clusters, presenting challenges for standard model-based stream clustering algorithms. This is the case of a dataset extracted from the Brazilian NF-e Project containing electronic invoice product descriptions, including many product clusters. While LDA-based clustering methods have shown to be crucial, they have been mainly evaluated on datasets with few clusters. We propose the Topic Model with Contextual Outlier Handling (TMCOH) method to overcome this limitation. This method combines the Dirichlet Process, specific word representation, and contextual outlier detection techniques to recycle identified outliers aiming to integrate them into appropriate clusters later on. The experimental results for our case study demonstrate the effectiveness of TMCOH when compared to state-of-the-art methods and its potential for application to text clustering in large datasets.

Topics discussed on social media platforms contain a disparate amount of information written in colloquial language, making it difficult to understand the narrative of the topic. In this paper, we take a step forward, towards the resolution of this problem by proposing a framework that performs the automatic extraction of narratives from a document, such as tweet posts. To this regard, we propose a methodology that extracts information from the texts through a pipeline of tasks, such as co-reference resolution and the extraction of entity relations. The result of this process is embedded into an annotation file to be used by subsequent operations, such as visualization schemas. We named this framework Tweet2Story and measured its effectiveness under an evaluation schema that involved three different aspects: (i) as an Open Information extraction (OpenIE) task, (ii) by comparing the narratives of manually annotated news articles linked to tweets about the same topic and (iii) by comparing their knowledge graphs, produced by the narratives, in a qualitative way. The results obtained show a high precision and a moderate recall, on par with other OpenIE state-of-the-art frameworks and confirm that the narratives can be extracted from small texts. Furthermore, we show that the narrative can be visualized in an easily understandable way.

Argumentation Mining (AM) is a growing sub-field of Natural Language Processing (NLP) which aims at extracting argumentative structures from text. In this work, neural learning and symbolic reasoning are combined in a system named N-SAUR, that extracts the argumentative structures present in a collection of texts, and then assesses each argument’s strength. The extraction is based on Toulmin’s model and the result quality surpasses previous approaches over an existing benchmark. Complementary scores are also extracted and combined with a set of rules that produce the final calculation of argument strength. The performance of the system was evaluated through human assessments. Users can also interact with the system in various ways, allowing for the strength calculation to change through user-cooperative reasoning.

Event extraction is an Information Retrieval task that commonly consists of identifying the central word for the event (trigger) and the event’s arguments. This task has been extensively studied for English but lags behind for Portuguese, partly due to the lack of task-specific annotated corpora. This paper proposes a framework in which two separated BERT-based models were fine-tuned to identify and classify events in Portuguese documents. We decompose this task into two sub-tasks. Firstly, we use a token classification model to detect event triggers. To extract event arguments, we train a Question Answering model that queries the triggers about their corresponding event argument roles. Given the lack of event annotated corpora in Portuguese, we translated the original version of the ACE-2005 dataset (a reference in the field) into Portuguese, producing a new corpus for Portuguese event extraction. To accomplish this, we developed an automatic translation pipeline. Our framework obtains F1 marks of 64.4 for trigger classification and 46.7 for argument classification setting, thus a new state of the art reference for these tasks in Portuguese.

Deep learning approaches have become popular in many different areas, including sentiment analysis (SA), because of their competitive performance. However, the downside of this approach is that they do not provide understandable explanations on how the sentiment values are calculated. In contrast, previous approaches that used sentiment lexicons can do that, but their performance is normally not high. To leverage the strengths of both approaches, we present a neuro-symbolic approach that combines deep learning (DL) and symbolic methods for SA tasks. The DL approach uses a pre-trained language model (PLM) to construct sentiment lexicon. The symbolic approach exploits the constructed sentiment lexicon and manually constructed shifter patterns to determine the sentiment of a sentence. Our experimental results show that the proposed approach leads to promising results with the additional advantage that sentiment predictions can be accompanied by understandable explanations.

The recent success of Large Language Models (LLMs) has sparked concerns about their potential to spread misinformation. As a result, there is a pressing need for tools to identify “fake arguments” generated by such models. To create these tools, examples of texts generated by LLMs are needed. This paper introduces a methodology to obtain good, bad and ugly arguments from argumentative essays produced by ChatGPT, OpenAI’s LLM. We then describe a novel dataset containing a set of diverse arguments, ArGPT. We assess the effectiveness of our dataset and establish baselines for several argumentation-related tasks. Finally, we show that the artificially generated data relates well to human argumentation and thus is useful as a tool to train and test systems for the defined tasks.

To advance the neural encoding of Portuguese (PT), and a fortiori the technological preparation of this language for the digital age, we developed a Transformer-based foundation model that sets a new state of the art in this respect for two of its variants, namely European Portuguese from Portugal (PT-PT) and American Portuguese from Brazil (PT-BR). To develop this encoder, which we named Albertina PT-*, a strong model was used as a starting point, DeBERTa, and its pre-training was done over data sets of Portuguese, namely over a data set we gathered for PT-PT and over the brWaC corpus for PT-BR. The performance of Albertina and competing models was assessed by evaluating them on prominent downstream language processing tasks adapted for Portuguese. Both Albertina versions are distributed free of charge and under a most permissive license possible and can be run on consumer-grade hardware, thus seeking to contribute to the advancement of research and innovation in language technology for Portuguese.

We describe OSPT, a new linguistic resource for European Portuguese that comprises more than 1.5 million Portuguese-Portuguese sentential paraphrase pairs. We generated the pairs automatically by using neural machine translation to translate the non-Portuguese side of a large parallel corpus. We hope this new corpus can be a valuable resource for paraphrase generation and provide a rich semantic knowledge source to improve downstream natural language understanding tasks. To show the quality and utility of such a dataset, we use it to train paraphrastic sentence embeddings and evaluate them in the ASSIN2 semantic textual similarity (STS) competition. We found that semantic embeddings trained on a small subset of OSPT can produce better semantic embeddings than the ones trained in the finely curated ASSIN2’s training data. Additionally, we show OSPT can be used for paraphrase generation with the potential to produce good data augmentation systems that pseudo-translate from Brazilian Portuguese to European Portuguese.

Dialogue systems need to deal with the unpredictability of user intents to track dialogue state and the heterogeneity of slots to understand user preferences. In this paper we investigate the hypothesis that solving these challenges as one unified model will allow the transfer of parameter support data across the different tasks. The proposed principled model is based on a Transformer encoder, trained on multiple tasks, and leveraged by a rich input that conditions the model on the target inferences. Conditioning the Transformer encoder on multiple target inferences over the same corpus, i.e., intent and multiple slot types, allows learning richer language interactions than a single-task model would be able to. In fact, experimental results demonstrate that conditioning the model on an increasing number of dialogue inference tasks leads to improved results: on the MultiWOZ dataset, the joint intent and slot detection can be improved by 3.2% by conditioning on intent, 10.8% by conditioning on slot and 14.4% by conditioning on both intent and slots. Moreover, on real conversations with Farfetch costumers, the proposed conditioned BERT can achieve high joint-goal and intent detection performance throughout a dialogue.

We tackle a single-machine scheduling problem where each job is characterized by weight, duration, due date, and deadline, while the objective is to minimize the weighted number of tardy jobs. The problem is strongly NP-hard and has practical applications in various domains, such as customer service and production planning. The best known exact approach uses a branch-and-bound structure, but its efficiency varies depending on the distribution of job parameters. To address this, we propose a new data-driven heuristic algorithm that considers the parameter distribution and uses machine learning and integer linear programming to improve the optimality gap. The algorithm also guarantees to obtain a feasible solution if it exists. Experimental results show that the proposed approach outperforms the current state-of-the-art heuristic.

Learning a well-informed heuristic function for hard planning domains is an elusive problem. Although there are known neural network architectures to represent such heuristic knowledge, it is not obvious what concrete information is learned and whether techniques aimed at understanding the structure help in improving the quality of the heuristics. This paper presents a network model that learns a heuristic function capable of relating distant parts of the state space via optimal plan imitation using the attention mechanism which drastically improves the learning of a good heuristic function. The learning of this heuristic function is further improved by the use of curriculum learning, where newly solved problem instances are added to the training set, which, in turn, helps to solve problems of higher complexities and train from harder problem instances. The methodologies used in this paper far exceed the performances of all existing baselines including known deep learning approaches and classical planning heuristics. We demonstrate its effectiveness and success on grid-type PDDL domains, namely Sokoban, maze-with-teleports and sliding tile puzzles.

The aim of this paper is to explore agent-based evacuation models in Python by conducting a systematic literature search using the PRISMA methodology. The principles of evacuation models are briefly described. Python packages and libraries for agent-based modelling frameworks are explained. Two research questions are defined. The first question aims to find out what typical current agent-based evacuation models look like in sense of application domain and location, number of agents, time and space scale etc.). The second question focuses on the details of the use of the Python programming language and libraries in implementations of agent-based evacuation models. The results of the PRISMA review are presented. Overall, Python is a suitable language for the development of agent-based evacuation models, as evidenced by the number of programming libraries and tools, as well as the growing number of scientific publications in last six years. However, most of the currently published models suffer from many shortcomings. A main surprise is the lack of adherence to standards in describing the agent-based computational model, providing source code and sharing documentation of experiments.