Search Results (126)

With the rapid growth of job-related crimes in developing economies, there is an urgent need for intelligent judicial systems to standardize sentencing practices. This study proposes a Multi-Level Feature Fusion Network (MLFFN) to enhance the accuracy and interpretability of sentencing predictions in job-related crime cases. The model integrates hierarchical legal feature representation, beginning with benchmark judgments (including starting-point penalties and additional penalties) as the foundational input. The frontend of MLFFN employs an attention mechanism to dynamically fuse word-level, segment-level, and position-level embeddings, generating a global feature encoding that captures critical legal relationships. The backend utilizes sliding-window convolutional kernels to extract localized features from the global feature map, preserving nuanced contextual factors that influence sentencing ranges. Trained on a dataset of job-related crime cases, MLFFN achieves a 6%+ performance improvement over the baseline models (BERT-base-Chinese, TextCNN, and ERNIE) in sentencing prediction accuracy. Our findings demonstrate that explicit modeling of legal hierarchies and contextual constraints significantly improves judicial AI systems. Full article

Despite rapid progress in natural language processing, current interactive AI systems continue to struggle with interpreting ambiguous, idiomatic, and contextually rich human language, a barrier to natural human–computer interaction. Many deployed applications, such as language games or educational tools, showcase surface-level adaptation but do not systematically probe or advance the deeper semantic understanding of user intent in open-ended, creative settings. In this paper, we present Semantrix, a web-based semantic word-guessing platform, not merely as a game but as a living testbed for evaluating and extending the semantic capabilities of state-of-the-art Transformer models in human-facing contexts. Semantrix challenges models to both assess the nuanced meaning of user guesses and generate dynamic, context-sensitive hints in real time, exposing the system to the diversity, ambiguity, and unpredictability of genuine human interaction. To empirically investigate how advanced semantic representations and adaptive language feedback affect user experience, we conducted a preregistered 2 × 2 factorial study (N = 42), independently manipulating embedding depth (Transformers vs. Word2Vec) and feedback adaptivity (dynamic hints vs. minimal feedback). Our findings revealed that only the combination of Transformer-based semantic modelling and adaptive hint generation sustained user engagement, motivation, and enjoyment; conditions lacking either component led to pronounced attrition, highlighting the limitations of shallow or static approaches. Beyond benchmarking game performance, we argue that the methodologies applied in platforms like Semantrix are helpful for improving machine understanding of natural language, paving the way for more robust, intuitive, and human-aligned AI approaches. Full article

Biological-sequence representation methods are pivotal for advancing machine learning in computational biology, transforming nucleotide and protein sequences into formats that enhance predictive modeling and downstream task performance. This review categorizes these methods into three developmental stages: computational-based, word embedding-based, and large language model (LLM)-based, detailing their principles, applications, and limitations. Computational-based methods, such as k-mer counting and position-specific scoring matrices (PSSM), extract statistical and evolutionary patterns to support tasks like motif discovery and protein–protein interaction prediction. Word embedding-based approaches, including Word2Vec and GloVe, capture contextual relationships, enabling robust sequence classification and regulatory element identification. Advanced LLM-based methods, leveraging Transformer architectures like ESM3 and RNAErnie, model long-range dependencies for RNA structure prediction and cross-modal analysis, achieving superior accuracy. However, challenges persist, including computational complexity, sensitivity to data quality, and limited interpretability of high-dimensional embeddings. Future directions prioritize integrating multimodal data (e.g., sequences, structures, and functional annotations), employing sparse attention mechanisms to enhance efficiency, and leveraging explainable AI to bridge embeddings with biological insights. These advancements promise transformative applications in drug discovery, disease prediction, and genomics, empowering computational biology with robust, interpretable tools. Full article

With the advancement of artificial intelligence, Agricultural Expert Systems (AESs) show great potential in enhancing agricultural management efficiency and resource utilization. Accurate extraction of semantic features from agricultural short texts is fundamental to enabling key functions such as intelligent question answering, semantic retrieval, and decision support. However, existing single-structure deep neural networks struggle to capture the hierarchical linguistic patterns and contextual dependencies inherent in domain-specific texts. To address this limitation, we propose a hybrid deep learning model—Bidirectional Encoder Recurrent Neural Network (BERNN)—which combines a domain-specific pre-trained Transformer encoder (AgQsBERT) with a Bidirectional Long Short-Term Memory (BiLSTM) network. AgQsBERT generates contextualized word embeddings by leveraging domain-specific pretraining, effectively capturing the semantics of agricultural terminology. These embeddings are then passed to the BiLSTM, which models sequential dependencies in both directions, enhancing the model’s understanding of contextual flow and word disambiguation. Importantly, the bidirectional nature of the BiLSTM introduces a form of architectural symmetry, allowing the model to process input in both forward and backward directions. This symmetric design enables balanced context modeling, which improves the understanding of fragmented and ambiguous phrases frequently encountered in agricultural texts. The synergy between semantic abstraction from AgQsBERT and symmetric contextual modeling from BiLSTM significantly enhances the expressiveness and generalizability of the model. Evaluated on a self-constructed agricultural question dataset with 110,647 annotated samples, BERNN achieved a classification accuracy of 97.19%, surpassing the baseline by 3.2%. Cross-domain validation on the Tsinghua News dataset further demonstrates its robust generalization capability. This architecture provides a powerful foundation for intelligent agricultural question-answering systems, semantic retrieval, and decision support within smart agriculture applications. Full article

Sentiment analysis methods for interactive services such as Danmu in online videos are challenged by their colloquial style and diverse sentiment expressions. For instance, the existing methods cannot easily distinguish between similar sentiments. To address these limitations, this paper proposes a dual-channel model integrated with attention mechanisms for multi-dimensional sentiment analysis of Danmu. First, we replace word embeddings with character embeddings to better capture the colloquial nature of Danmu text. Second, the dual-channel multi-dimensional sentiment encoder extracts both the high-level semantic and raw contextual information. Channel I of the encoder learns the sentiment features from different perspectives through a mixed model that combines the benefits of self-Attention and Dilated CNN (ADCMix) and performs contextual modeling through bidirectional long short-term memory (BiLSTM) with attention mechanisms. Channel II mitigates potential biases and omissions in the sentiment features. The model combines the two channels to erase the fuzzy boundaries between similar sentiments. Third, a multi-dimensional sentiment decoder is designed to handle the diversity in sentiment expressions. The superior performance of the proposed model is experimentally demonstrated on two datasets. Our model outperformed the state-of-the-art methods on both datasets, with improvements of at least 2.05% in accuracy and 3.28% in F1-score. Full article

Dementia is a neurodegenerative disorder characterized by progressive cognitive impairment that significantly affects daily living. Early detection of Alzheimer’s disease—the most common form of dementia—remains essential for prompt intervention and treatment, yet clinical diagnosis often requires extensive and resource-intensive procedures. This article explores the effectiveness of automated Natural Language Processing (NLP) methods for identifying Alzheimer’s indicators from audio transcriptions of the Cookie Theft picture description task in the PittCorpus dementia database. Five NLP approaches were compared: a classical Tf–Idf statistical representation and embeddings derived from large language models (GloVe, BERT, Gemma-2B, and Linq-Embed-Mistral), each integrated with a logistic regression classifier. Transcriptions were carefully preprocessed to preserve linguistically relevant features such as repetitions, self-corrections, and pauses. To compare the performance of the five approaches, a stratified 5-fold cross-validation was conducted; the best results were obtained with BERT embeddings (84.73% accuracy) closely followed by the simpler Tf–Idf approach (83.73% accuracy) and the state-of-the-art model Linq-Embed-Mistral (83.54% accuracy), while Gemma-2B and GloVe embeddings yielded slightly lower performances (80.91% and 78.11% accuracy, respectively). Contrary to initial expectations—that richer semantic and contextual embeddings would substantially outperform simpler frequency-based methods—the competitive accuracy of Tf–Idf suggests that the choice and frequency of the words used might be more important than semantic or contextual information in Alzheimer’s detection. This work represents an effort toward implementing user-friendly software capable of offering an initial indicator of Alzheimer’s risk, potentially reducing the need for an in-person clinical visit. Full article

The Adaptive Skip-gram (AdaGram) algorithm extends traditional word embeddings by learning multiple vector representations per word, enabling the capture of contextual meanings and polysemy. Originally implemented in Julia, AdaGram has seen limited adoption due to ecosystem fragmentation and the comparative scarcity of Julia’s machine learning tooling compared to Python’s mature frameworks. In this work, we present a Python-based reimplementation of AdaGram that facilitates broader integration with modern machine learning tools. Our implementation expands the model’s applicability beyond natural language, enabling the analysis of scientific notation—particularly chemical and physical formulas encoded in LaTeX. We detail the algorithmic foundations, preprocessing pipeline, and hyperparameter configurations needed for interdisciplinary corpora. Evaluations on real-world texts and LaTeX-encoded formulas demonstrate AdaGram’s effectiveness in unsupervised word sense disambiguation. Comparative analyses highlight the importance of corpus design and parameter tuning. This implementation opens new applications in formula-aware literature search engines, ambiguity reduction in automated scientific summarization, and cross-disciplinary concept alignment. Full article

Aspect-Based Sentiment Analysis (ABSA) has gained significant popularity in recent years, which emphasizes the aspect-level sentiment representation of sentences. Current methods for ABSA often use pre-trained models and graph convolution to represent word dependencies. However, they struggle with long-range dependency issues in lengthy texts, resulting in averaging and loss of contextual semantic information. In this paper, we explore how richer semantic relationships can be encoded more efficiently. Inspired by quantum theory, we construct superposition states from text sequences and utilize them with quantum measurements to explicitly capture complex semantic relationships within word sequences. Specifically, we propose an attention-based semantic dependency fusion method for ABSA, which employs a quantum embedding module to create a superposition state of real-valued word sequence features in a complex-valued Hilbert space. This approach yields a word sequence density matrix representation that enhances the handling of long-range dependencies. Furthermore, we introduce a quantum cross-attention mechanism to integrate sequence features with dependency relationships between specific word pairs, aiming to capture the associations between particular aspects and comments more comprehensively. Our experiments on the SemEval-2014 and Twitter datasets demonstrate the effectiveness of the quantum-inspired attention-based semantic dependency fusion model for the ABSA task. Full article

Background/Objectives: Sleep disorders are prevalent among aging populations and are often linked to cognitive decline, chronic conditions, and reduced quality of life. Traditional diagnostic methods, such as polysomnography, are resource-intensive and limited in accessibility. Meanwhile, individuals frequently describe their sleep experiences through unstructured narratives in clinical notes, online forums, and telehealth platforms. This study proposes a machine learning pipeline (sleepCare) that classifies sleep-related narratives into clinically meaningful categories, including stress-related, neurodegenerative, and breathing-related disorders. The proposed framework employs natural language processing (NLP) and machine learning techniques to support remote applications and real-time patient monitoring, offering a scalable solution for the early identification of sleep disturbances. Methods: The sleepCare consists of a three-tiered classification pipeline to analyze narrative sleep reports. First, a baseline model used a Multinomial Naïve Bayes classifier with n-gram features from a Bag-of-Words representation. Next, a Support Vector Machine (SVM) was trained on GloVe-based word embeddings to capture semantic context. Finally, a transformer-based model (BERT) was fine-tuned to extract contextual embeddings, using the [CLS] token as input for SVM classification. Each model was evaluated using stratified train-test splits and 10-fold cross-validation. Hyperparameter tuning via GridSearchCV optimized performance. The dataset contained 475 labeled sleep narratives, classified into five etiological categories relevant for clinical interpretation. Results: The transformer-based model utilizing BERT embeddings and an optimized Support Vector Machine classifier achieved an overall accuracy of 81% on the test set. Class-wise F1-scores ranged from 0.72 to 0.91, with the highest performance observed in classifying normal or improved sleep (F1 = 0.91). The macro average F1-score was 0.78, indicating balanced performance across all categories. GridSearchCV identified the optimal SVM parameters (C = 4, kernel = ‘rbf’, gamma = 0.01, degree = 2, class_weight = ‘balanced’). The confusion matrix revealed robust classification with limited misclassifications, particularly between overlapping symptom categories such as stress-related and neurodegenerative sleep disturbances. Conclusions: Unlike generic large language model applications, our approach emphasizes the personalized identification of sleep symptomatology through targeted classification of the narrative input. By integrating structured learning with contextual embeddings, the framework offers a clinically meaningful, scalable solution for early detection and differentiation of sleep disorders in diverse, real-world, and remote settings. Full article

The rise in social media has improved communication but also amplified the spread of hate speech, creating serious societal risks. Automated detection remains difficult due to subjectivity, linguistic diversity, and implicit language. While prior research focuses on high-resource languages, this study addresses the underexplored multilingual challenges of Arabic and Urdu hate speech through a comprehensive approach. To achieve this objective, this study makes four different key contributions. First, we have created a unique multi-lingual, manually annotated binary and multi-class dataset (UA-HSD-2025) sourced from X, which contains the five most important multi-class categories of hate speech. Secondly, we created detailed annotation guidelines to make a robust and perfect hate speech dataset. Third, we explore two strategies to address the challenges of multilingual data: a joint multilingual and translation-based approach. The translation-based approach involves converting all input text into a single target language before applying a classifier. In contrast, the joint multilingual approach employs a unified model trained to handle multiple languages simultaneously, enabling it to classify text across different languages without translation. Finally, we have employed state-of-the-art 54 different experiments using different machine learning using TF-IDF, deep learning using advanced pre-trained word embeddings such as FastText and Glove, and pre-trained language-based models using advanced contextual embeddings. Based on the analysis of the results, our language-based model (XLM-R) outperformed traditional supervised learning approaches, achieving 0.99 accuracy in binary classification for Arabic, Urdu, and joint-multilingual datasets, and 0.95, 0.94, and 0.94 accuracy in multi-class classification for joint-multilingual, Arabic, and Urdu datasets, respectively. Full article

We present a new framework to study the stability of semantic projections based on word embeddings. Roughly speaking, semantic projections are indices taking values in the interval $[0,1]$ that measure how terms share contextual meaning with the words of a given universe. Since there are many ways to define such projections, it is important to establish a procedure for verifying whether a group of them behaves similarly. Moreover, when fixing one particular projection, it is important to assess whether the average projections remain consistent when replacing the original universe with a similar one describing the same semantic environment. The aim of this paper is to address the lack of formal tools for assessing the stability of semantic projections (that is, their invariance under formal changes which preserve the underlying semantic context) across alternative but semantically related universes in word embedding models. To address these problems, we employ a combination of statistical and AI methods, including correlation analysis, clustering, chi-squared distance measures, weighted approximations, and Lipschitz-based estimators. The methodology provides theoretical guarantees under mild mathematical assumptions, ensuring bounded errors in projection estimations based on the assumption of Lipschitz continuity. We demonstrate the practical applicability of our approach through two case studies involving agricultural terminology across multiple data sources (DOAJ, Scholar, Google, and Arxiv). Our results show that semantic stability can be quantitatively evaluated and that the careful modeling of projection functions and universes is crucial for robust semantic analysis in NLP. Full article

Large language models (LLMs) have made significant strides in generating coherent and contextually relevant responses across diverse domains. However, these advancements have also led to an increase in adversarial attacks, such as prompt injection, where attackers embed malicious instructions within prompts to bypass security filters and manipulate LLM outputs. Various injection techniques, including masking and encoding sensitive words, have been employed to circumvent security measures. While LLMs continuously enhance their security protocols, they remain vulnerable, particularly in multimodal contexts. This study introduces a novel method for bypassing LLM security policies by embedding malicious instructions within a mind map image. The attack leverages the intentional incompleteness of the mind map structure, specifically the absence of explanatory details. When the LLM processes the image and fills in the missing sections, it inadvertently generates unauthorized outputs, violating its intended security constraints. This approach applies to any LLM capable of extracting and interpreting text from images. Compared to the best-performing baseline method, which achieved an ASR of 30.5%, our method reaches an ASR of 90%, yielding an approximately threefold-higher attack success. Understanding this vulnerability is crucial for strengthening security policies in state-of-the-art LLMs. Full article

In recent years, text classification has found wide application in diverse real-world scenarios. In Chinese news classification tasks, limitations such as sparse contextual information and semantic ambiguity exist in the title text. To improve the performance of short text classification, this paper proposes a Word2Vec-based enhanced word embedding method and exhibits the design of a dual-channel hybrid neural network architecture to effectively extract semantic features. Specifically, we introduce a novel weighting scheme, Term Frequency-Document Frequency Category-Distribution Weight (TF-IDF-CDW), where Category Distribution Weight (CDW) reflects the distribution pattern of words across different categories. By weighting the pretrained Word2Vec vectors with TF-IDF-CDW and concatenating them with part-of-speech (POS) feature vectors, semantically enriched and more discriminative word embedding vectors are generated. Furthermore, we propose a dual-channel hybrid model based on a Gated Convolutional Neural Network (GCNN) and Bidirectional Long Short-Term Memory (BiLSTM), which jointly captures local features and long-range global dependencies. To evaluate the overall performance of the model, experiments were conducted on the Chinese short text datasets THUCNews and TNews. The proposed model achieved classification accuracies of 91.85% and 87.70%, respectively, outperforming several comparative models and demonstrating the effectiveness of the proposed method. Full article

Detecting hate speech in social media is challenging due to its rarity, high-dimensional complexity, and implicit expression via sarcasm or spelling variations, rendering linear models ineffective. In this study, the SVM (Support Vector Machine) algorithm is used to map text features from low-dimensional to high-dimensional space using kernel function techniques to meet complex nonlinear classification challenges. By maximizing the category interval to locate the optimal hyperplane and combining nuclear techniques to implicitly adjust the data distribution, the classification accuracy of hate speech detection is significantly improved. Data collection leverages social media APIs (Application Programming Interface) and customized crawlers with OAuth2.0 authentication and keyword filtering, ensuring relevance. Regular expressions validate data integrity, followed by preprocessing steps such as denoising, stop-word removal, and spelling correction. Word embeddings are generated using Word2Vec’s Skip-gram model, combined with TF-IDF (Term Frequency–Inverse Document Frequency) weighting to capture contextual semantics. A multi-level feature extraction framework integrates sentiment analysis via lexicon-based methods and BERT for advanced sentiment recognition. Experimental evaluations on two datasets demonstrate the SVM model’s effectiveness, achieving accuracies of 90.42% and 92.84%, recall rates of 88.06% and 90.79%, and average inference times of 3.71 ms and 2.96 ms. These results highlight the model’s ability to detect implicit hate speech accurately and efficiently, supporting real-time monitoring. This research contributes to creating a safer online environment by advancing hate speech detection methodologies. Full article

Sentiment analysis in Chinese microblogs is challenged by complex syntactic structures and fine-grained sentiment shifts. To address these challenges, a Contextually Enriched Graph Neural Network (CE-GNN) is proposed, integrating self-supervised learning, context-aware sentiment embeddings, and Graph Neural Networks (GNNs) to enhance sentiment classification. First, CE-GNN is pre-trained on a large corpus of unlabeled text through self-supervised learning, where Masked Language Modeling (MLM) and Next Sentence Prediction (NSP) are leveraged to obtain contextualized embeddings. These embeddings are then refined through a context-aware sentiment embedding layer, which is dynamically adjusted based on the surrounding text to improve sentiment sensitivity. Next, syntactic dependencies are captured by Graph Neural Networks (GNNs), where words are represented as nodes and syntactic relationships are denoted as edges. Through this graph-based structure, complex sentence structures, particularly in Chinese, can be interpreted more effectively. Finally, the model is fine-tuned on a labeled dataset, achieving state-of-the-art performance in sentiment classification. Experimental results demonstrate that CE-GNN achieves superior accuracy, with a Macro F-measure of 80.21% and a Micro F-measure of 82.93%. Ablation studies further confirm that each module contributes significantly to the overall performance. Full article