Search

Text simplification reduces the complexity of text while preserving essential information, thus making it more accessible to a broad range of readers, including individuals with cognitive disorders, non-native speakers, children, and the general public. In this paper, we present experiments on text simplification for the Lithuanian language, aiming to simplify administrative texts to a Plain Language level. We fine-tuned mT5 and mBART models for this task and evaluated the effectiveness of ChatGPT as well. We assessed simplification results via both quantitative metrics and qualitative evaluation. Our findings indicated that mBART performed the best as it achieved the best scores across all evaluation metrics. The qualitative analysis further supported these findings. ChatGPT experiments showed that it responded quite well to a short and simple prompt to simplify the given text; however, it ignored most of the rules given in a more elaborate prompt. Finally, our analysis revealed that BERTScore and ROUGE aligned moderately well with human evaluations, while BLEU and readability scores indicated lower or even negative correlations [ABSTRACT FROM AUTHOR]

Imbalanced datasets are one of the main issues encountered by artificial intelligence researchers, as machine learning (ML) algorithms can become biased toward the majority class and perform insufficiently on the minority classes. Financial distress (FD) is one of the numerous real-world applications of ML, struggling with this issue. Furthermore, the topic of financial distress holds considerable interest for both academics and practitioners due to the non-determined indicators of condition states. This research focuses on the involvement of balancing techniques according to different FD condition states. Moreover, this research was expanded by implementing ML models and dimensionality reduction techniques. During the course of this study, a Combined FD was constructed using five distinct conditions, ten distinct class balancing techniques, five distinct dimensionality reduction techniques, two features selection strategies, eleven machine learning models, and twelve weighted majority algorithms (WMAs). Results revealed that the highest area under the receiver operating characteristic (ROC) curve (AUC) score was achieved when using the extreme gradient boosting machine (XGBoost) feature selection technique, the experimental max number strategy, the undersampling methods, and the WMA 3.1 weighted majority algorithm (i.e., with categorical boosting (CatBoost), XGBoost, and random forest (RF) having equal voting weights). Moreover, this research has introduced a novel approach for setting the condition states of financial distress, including perspectives from debt and change in employment. These outcomes have been achieved utilizing authentic enterprise data from small and medium Lithuanian enterprises. [ABSTRACT FROM AUTHOR]

According to the Transport Innovation Association estimations, on average, a single lorry driver carries about 50 sheets of paper consisting of only 15 CMR documents. In Lithuania alone, there are about 50,000 active trucks each month, resulting in about 192 tonnes of wasted paper annually. The manual entry of document data into enterprise resource planning (ERP) systems not only is time-consuming but inefficient and could consist of errors. To address these issues, a framework for the digitisation of logistic documents, such as invoices, receipts and CMRs, is proposed that uses object detection, optical character recognition (OCR) and a semi-supervised finetuning pipeline. This study focuses on both the experimentation and implementation phases of research. During the experimentation phase, multiple object detection models like SSD MobileNet, SSD ResNet-50, Faster-RCNN, EfficientDet-D4, and CenterNet HourGlass104 were evaluated. OCR models like Tesseract, EasyOCR, KerasOCR, Kraken, Doctr, and Google OCR were tested. Extensive evaluations showed that using the combination of Faster-RCNN and Google OCR works the best for document digitisation. The object detection model was trained using approximately 1000 images that were equally distributed among three classes of documents. The Faster-RCNN model achieved an average precision (AP) of 0.95 at an Intersection over the Union (IoU) threshold of 0.5, 0.84 AP at an IoU of 0.75, and 0.71 AP IoU ranging from 0.5 to 0.95, with an average recall (AR) of 0.75, within the same range. OCRs performances were manually assessed due to the lack of annotations, with Google OCR proving the best results in the presence of minor inaccuracies in bounding box placement or noise within the bounding box. To further increase the accuracy of the object detection model, a semi-automated labelling process was introduced, where a trained Faster-RCNN model is used to generate initial bounding boxes and class labels on unseen data, which later are manually adjusted for further finetuning of a pre-trained Faster-RCNN model. The proposed system is an improvement in automating logistics document digitisation, reducing dependence on manual labour and an overall increase of efficiency in the transportation industry. [ABSTRACT FROM AUTHOR]

The precise identification of enterprise activity codes stands as a crucial task enabling the rapid and effective establishment or renewal of databases encompassing both public and private companies, which in return helps to make an informative decision about countries' economic tendencies. The research involves combining multi-source datasets, data cleaning, explanatory data analysis, retrieval of embeddings, feature selection, the optimal number of clusters identification, data clustering, and post-clustering analysis. Gathered insights allow for informative decisions about taxes, needed state aid and competition analysis. In both the Republic of Lithuania and the European Union, the enterprise classification system operates under the Nomenclature of Economic Activities (NACE), which employs a six-digit framework. For instance, code 461900 indicates that the business conducts the sales of various goods that involve agents. The initial two digits represent overarching enterprise classifications, in this case, retail trade, while the final four digits delineate specific categorisations within the country's industries. This study aims to apply clustering methods to help in the identification of the economic activities of enterprises using descriptions that could be found in the "Company Description" section of the rekvizitai.lt website. The dataset consists of 28350 business descriptions. Two main themes were observed in the data: (1) the average description lengths are 14, excluding stop-words; (2) the most common activities in the Lithuania economic sector are wholesale, retail, agriculture, and service industry. In this study, 3 embedding methods (BERT, LaBSE and Word2Vec), 4 feature selection methods (PCA, UMAP, SVD, and autoencoders) and 8 clustering methods (K-means, GMM, agglomerative, mean shift, OPTICS, BIRCH, HDBSCAN, DEC) were used for experimentations with 195 models trained in total. Three main metrics, silhouette score, Davies Bouldin score, and Calinski-Harabasz Index, are evaluated across all clustering algorithms, with adjusted Rand Index and mutual information evaluated for hard-clustering methods. The initial experiments showed that LaBSE and Word2Vec are the most prominent methods for embedding retrieval, while PCA and UMAP are most suitable for dimensionality reduction. The elbow approach was employed in additional experiments to determine the ideal number of clusters. Although these experiments demonstrated that data may be grouped into fewer clusters, the outcomes did not indicate a statistically significant improvement, and adhering to the original NACE space facilitates a more accurate assessment of the current economic landscape situation. Clustering results from K-means, agglomerative, and mean shift methods showed good intra-clustering and slightly above average inter-clustering results. This research demonstrates that enterprise activity sectors can be categorised using Lithuanian descriptions and the K-means, agglomerative, or mean shift clustering algorithms. Future research will focus on all three algorithms hyperparameter optimisation to improve inter-clustering and intra-clustering results. [ABSTRACT FROM AUTHOR]

Monitoring social and traditional media requires handling vast amounts of unstructured data, which can be challenging to manage and analyse effectively In such scenarios, techniques that can extract only key information, reducing the data volume and organising it into structured formats, prove to be highly beneficial. The focus of this study was on testing such method, based on the approach of Knowledge Graphs. Knowledge Graphs are a structured way of representing information by connecting entities through relationships, often captured using SVO (Subject-Verb-Object) triplets. These triplets allow for a clearer understanding of how different concepts are related within the analysed data. Using Knowledge Graphs, the occurrences of disinformation in the messaging platform Telegram were investigated. Disinformation was sought by calculating different Graph Kernels between Knowledge Graphs constructed from collected Telegram messages and graphs made from already confirmed disinformation cases published in the EUvsDisinfo database. This method was applied to analyse more than 1 million messages from 30 Russian and Belarusian Telegram channels selected by experts, which, as the initial analysis showed, are also followed by Lithuanian citizens and frequently contain manifestations of disinformation. The disinformation cases were specifically selected only when they were related to the ongoing war between Russia and Ukraine. The results of the analysis showed that a certain amount of disinformation appeared in all the analysed channels, but most channels particularly emphasised disinformation related to biological and nuclear weapons, as well as the views of the residents of the occupied territories. When analysing the methods, it was found that the best way to form Graph Kernels for disinformation detection was the Shortest Path Kernel. This method, unlike others, allowed distinguishing the graphs most characterised by disinformation from the large number of graphs, which was the primary goal of this study. [ABSTRACT FROM AUTHOR]

The rapid increase of online content, which is often coupled with the ease with which people can share their opinions, has contributed to a rise in social issues such as cyberbullying, insults, and hate speech. To mitigate these issues, some online platforms have implemented measures like disabling anonymous comments or completely removing the option to comment on articles the users used to have. Additionally, certain platforms employ human moderators to identify and remove hate speech. However, due to a huge volume of online interactions, manually moderating content requires substantial human resources. Advances in artificial intelligence, particularly in natural language processing (NLP), offer promising results in hate speech identification. Automated hate speech detection systems can facilitate content moderation by effi- ciently processing and managing large volumes of data. In this study, we present a comparative evaluation of hate speech detection solutions for the Lithuanian language. We used several deep learning models for hate speech detection: Multilingual BERT, LitLat BERT, Electra, open Llama2 for the Lithuanian language, RWKV, BiLSTM, LSTM, CNN and ChatGPT. For the Electra model, we trained ourselves from scratch with Lithuanian texts that made more than 2.5 billion tokens. Multilingual BERT, LitLat BERT, Electra and RWKV were further fine-tuned to classify Lithuanian user-generated comments into three main classes: hate, offensive, and neutral speech. For comparison purposes, we also trained BiLSTM, LSTM and CNN models for the task. Open Llama2 for the Lithuanian language and ChatGPT were used without fine-tuning, and Open Llama2 for the Lithuanian language was then fine-tuned to get better results. To train or adapt the models to the hate speech detection task, we prepared an annotated dataset. It has had 27 357 user-generated comments (hate speech -- 4220, offensive -- 7821, neutral -- 15 316). All models were evaluated with accuracy, precision, recall, and F1-score metrics. Our future plans include augmentation of our annotated dataset with additional data sources and hate topics as well as experiments in model bias, robustness and output explainability. [ABSTRACT FROM AUTHOR]

Forest management in Lithuania is focused on sustainable development, balancing economic, environmental, and social objectives. Lithuanian forests, covering approximately one-third of the country's land area, are a critical resource for biodiversity, timber production, carbon sequestration, and recreation. To better understand how forests develop over time and how various factors influence their growth, it is important to identify tree growth functions. These functions are mathematical models that describe the relationship between forest growth and other variables such as age, environmental conditions, and management practices. Additionally, AI plays a crucial role in forest planning and management due to its ability to process large datasets, enhance decision-making, and improve sustainability practices. AI and machine learning (ML) are extensively used in forestry, as these methods enable the analysis of large datasets, including climate data, soil conditions, and historical forest management records, to optimise forest operations. For example, AI algorithms can help plan sustainable logging activities by predicting tree growth patterns and yield, while ML models assess the impact of different forest management practices on biodiversity and ecosystem health. AI-driven decision support systems can also assist in the realtime monitoring and management of forest resources, ensuring efficient and sustainable practices. This research aims to develop tree growth functions for Lithuanian forests and extend the GAYA stand simulator to new contexts, adapting it to Lithuanian conditions and reflecting the country's unique compositions of species and ecological conditions. We are integrating these growth functions into the GAYA framework to ensure accurate projections for Lithuanian forest ecosystems. Moreover, through a literature review, we have identified ways to improve forest planning using AI methods. Initial results show that estimating tree volume growth requires accounting for various forest parameters such as volume, height, age, diameter, and information about cutting activities. By applying ML models, forest planning can be further improved by integrating diverse datasets and evaluating multiple forest management scenarios. [ABSTRACT FROM AUTHOR]

Cancer patients often experience fluctuations in their health, which can be challenging to monitor continuously through traditional methods, such as patient self-reporting during clinic visits. With the growing prevalence of smartphones and their built-in sensors, new opportunities have emerged for real-time, passive monitoring of patient behaviour and health. These phone sensors -- tracking factors like physical activity, location, and phone usage -- can provide continuous data that may signal subtle changes in a patient's condition even before they become consciously aware of worsening symptoms. This research explores the use of phone sensor data to understand how real-time behavioural changes might be related to the onset of symptoms in cancer patients. By leveraging sensor data alongside patient surveys, the study investigates how patterns in daily activity, such as movement and phone usage, might signal an impending change in symptoms. The study included patients grouped by cancer type, gender, functional status (ECOG scale), and age. Loess regression and statistical analysis were applied to examine the dynamics of sensor data before and after symptom onset. Various variables describing patients' activity, such as distance from home, screen time ratio, and others, were evaluated over time using phone sensors. The study's results revealed that patients' activity levels, or time spent at home, begin to decline even before the patient reports the onset of symptoms, such as fatigue. The study found that tracking sensor readings can help predict worsening symptoms and facilitate early intervention, which is crucial to the quality of patient care. These findings highlight the importance of using phone sensors for real-time patient monitoring, enabling personalised care for cancer patients. [ABSTRACT FROM AUTHOR]

Fake news is becoming a recognised problem in society, creating large volumes of misinformation and raising questions about media integrity. The spread of misinformation has serious consequences, affecting public opinion, undermining trust in institutions, and sometimes leading to real-world harm. The sheer volume of information produced daily making manual verification of news impractical and time-consuming, leading in the need for automated tools that can assist with the issue. In this work, we present a proof of concept of a tool designed to detect disinformation in English-language media sources. At the core of our solution is the RoBERTa model, fine-tuned on a diverse set of articles from American (mostly for non-disinformation) and Russian (mostly for disinformation) English-language sources. This approach allows us to save time by avoiding the need to train the model from scratch. Additionally, RoBERTa's advanced capabilities, such as its ability to grasp the context and meaning of complex sentences and its bidirectional nature (analysing sentences from both the beginning to the end and in the opposite way), enable it to capture long-range dependencies between words. These features are valuable in identifying complex linguistic structures, such as hyperbole, unverified claims, biased language, and sarcasm, commonly found in news sources. Besides, we are using this approach to collect and analyse texts from Lithuanian media sources, focusing on various domains such as politics, economy, society, and business, to identify the prevalence of disinformation within these outlets. This allows us to gain insights into how misinformation is distributed across different sectors. To enhance the usability of our solution, we developed a Dash library-based web application that displays the model's evaluation results and delivers an intuitive interface for users to interact with the system. This application allows users to upload news articles or plain text, analyse them for potential misinformation, and view detailed, real-time feedback on the model's predictions. Additionally, the system allows for batch processing of multiple texts at once, providing scalability for larger datasets. The result of our work is a developed and tested version of a deep learning based disinformation detection system, capable of analysing disinformation in the selected national and international news and media sources and presenting the analysis results with an informative dashboard. [ABSTRACT FROM AUTHOR]

Arun Kumar Mishra wrote that every movement of goods from one point to the next must have the attached documents. According to Statista, logistics industry worldwide grows 0.5 trillion dollars each year, meaning more transportation is required the more documents needs to be processed. To efficiently manage huge volumes of documents and automate a decision-making process a classification system is required. This research focuses on logistics documents classification utilizing deep learning and machine learning algorithms. For this study, 50 GB of unlabeled data were presented, and the initial experiments were conducted using 5078 manually selected documents. Manually selected documents were assigned to 4 commonly used logistics document categories: CMRs, invoices, receipts, and others. The dataset was split into train and test sets, where 80% or 4058 of the documents were designated for training and 20% or 1014 of the documents for testing. Five main preprocessing steps were applied: convertion from PDF to JPG, resizing, deskewing, tint and noise removal. Two main methodologies were applied, application of neural networks and traditional machine learning classification techniques. Both approaches utilized pretrained backbone models on ImageNet. For neural networks we used Efficient-Net80, VGG16, MobileNet, ResNet50, DenseNet and InceptionV3. The neural network with the ResNet50 backbone outperformed other models achieving 0.9582 accuracy, 0.9593 precision, 0.9582 recall and 0.9585 F1 score. Rest models showed comparable results in performance evaluation: EfficientNet80 achieved 0.9467, VGG16 0.9176, MobileNet 0.9307, DenseNet 0.9387 and InceptionV3 0.9387 F1 scores. In addition, traditional machine learning classifiers, including Support Vector Machines (SVM), Random Forest, K-Nearest Neighbors (KNN), and XGBoost (XGB), were trained using features extracted from the ResNet50 backbone. The best-performing machine learning model was the Support Vector Classifier, achieving an accuracy of 0.9471, 0.9470 precision, 0.9471 recall and 0.9466 F1 score, while the XGBoost classifier, Random Forest, and KNearest Neighbors classifiers achieved F1 scores of 0.9440, 0.9455, and 0.9281, respectively. The research showed that the most promising solution for logistics document classification is the ResNet50 model and that it could be implemented in logistic environments to automate document separation. Future research will focus on dataset expansion utilizing pretrained ResNet50 model to label the remaining unused documents and further fine-tune models to enhance model F1 score, minimizing the need for human intervention in document classification. [ABSTRACT FROM AUTHOR]