A review of sentiment analysis: tasks, applications, and deep learning techniques

International Journal of Data Science and Analytics, Год журнала: 2024, Номер unknown

Опубликована: Июль 1, 2024

Язык: Английский

---

Exploring sentiment analysis in handwritten and E-text documents using advanced machine learning techniques: a novel approach

Journal Of Big Data, Год журнала: 2025, Номер 12(1)

Опубликована: Янв. 23, 2025

Язык: Английский

---

Hybrid Deep Learning Model to Predict Students’ Sentiments in Higher Educational Institutions

International Journal of Interactive Mobile Technologies (iJIM), Год журнала: 2025, Номер 19(01), С. 46 - 61

Опубликована: Янв. 13, 2025

Sentiment analysis has been widely used in various fields of social media, education, and business. Specifically, the education domain, usage sentiment is difficult due to huge amount information, nature language, processing diverse perceptions students. Deep learning emerges as an advanced concept realm machine that learns features automatically from raw text data, making them well-suited for tasks. In recent years, deep analyzing sentiments. architectures have surpassed other paradigms performing analysis. The ability analyze students’ sentiments enables HEI process amounts unstructured data quickly, efficiently, cost-effectively. paper aims predict reviews posted VLE regarding online educators optimize their teaching methods best results. This study explores CNN, LSTM, hybrid CNN-LSTM prediction proposed architecture achieves superior performance compared baseline algorithms with respect accuracy, precision, recall, F1 score. According outcomes, recommended technique remarkable accuracy 97%. findings facilitate progress a more efficient system gives valuable insights volume textual data.

Язык: Английский

---

Low-resource MobileBERT for emotion recognition in imbalanced text datasets mitigating challenges with limited resources

PLoS ONE, Год журнала: 2025, Номер 20(1), С. e0312867 - e0312867

Опубликована: Янв. 24, 2025

Modern dialogue systems rely on emotion recognition in conversation (ERC) as a core element enabling empathetic and human-like interactions. However, the weak correlation between emotions semantics poses significant challenges to dialogue. Semantically similar utterances can express different types of emotions, depending context or speaker. In order tackle this challenge, our paper proposes novel loss called Focal Weighted Loss (FWL) with adversarial training compact language model MobileBERT. Our proposed function handles problem imbalanced classification through does not require large batch sizes more computational resources. approach has been employed four text benchmark datasets, MELD, EmoryNLP, DailyDialog IEMOCAP demonstrating competitive performance. Extensive experiments these datasets validate effectiveness FWL training. This enables interactions digital platforms. shows its potential deliver performance under limited resource constraints, comparable models.

Язык: Английский

---

Discovering Causal Relationships in Noisy Web Data for Sentiment Classification Using Attention Mechanisms

Lecture notes in computer science, Год журнала: 2025, Номер unknown, С. 357 - 377

Опубликована: Янв. 1, 2025

Язык: Английский

---

Explainable physics-guided attention network for long-lead ENSO forecasts

Information Sciences, Год журнала: 2025, Номер unknown, С. 122084 - 122084

Опубликована: Март 1, 2025

Язык: Английский

---

Text-Based Sentiment Analysis: Experimentation with CNN and Basic Machine-Learning Approaches

Smart innovation, systems and technologies, Год журнала: 2025, Номер unknown, С. 147 - 160

Опубликована: Янв. 1, 2025

Язык: Английский

---

Dynamic Scene Segmentation and Sentiment Analysis for Danmaku

Applied Sciences, Год журнала: 2025, Номер 15(8), С. 4435 - 4435

Опубликована: Апрель 17, 2025

Danmaku analysis is important for understanding video content and user interactions. However, current methods often look at separate comments do not see the complex links between video’s context. This paper presents a new approach that combines advanced shot segmentation techniques, using Deep Convolutional Neural Networks (DDCNN), with an of feelings based on MacBERT model. First, videos are cut into clear scenes detected scene changes. Then, large group collected studied to make complete dictionary this field. With as base, Danmaku-E model made find seven different emotional categories within comments. The shows significantly improved performance, accuracy increasing from 94.58% 95.37% F1 score going 94.92% 95.66%, helped by feelings. Experimental results show good effects expanded in helping performance structures. Also, Apriori algorithm used explain content, providing deeper participation reactions.

Язык: Английский

---

Investigating Stock Prediction Using LSTM Networks and Sentiment Analysis of Tweets Under High Uncertainty: A Case Study of North American and European Banks

IEEE Access, Год журнала: 2024, Номер 12, С. 122239 - 122248

Опубликована: Янв. 1, 2024

Язык: Английский

---

Automatic Extraction and Cluster Analysis of Natural Disaster Metadata Based on the Unified Metadata Framework

ISPRS International Journal of Geo-Information, Год журнала: 2024, Номер 13(6), С. 201 - 201

Опубликована: Июнь 14, 2024

The development of information technology has led to massive, multidimensional, and heterogeneously sourced disaster data. However, there’s currently no universal metadata standard for managing natural disasters. Common pre-training models extraction requiring extensive training data show somewhat limited effectiveness, with annotated resources. This study establishes a unified standard, utilizes self-trained (UIE) Python libraries extract stored in both structured unstructured forms, analyzes the results using Word2vec-Kmeans cluster algorithm. that (1) UIE model, learning rate 3 × 10−4 batch_size 32, significantly improves various disasters by over 50%. Our optimized model outperforms many other methods terms precision, recall, F1 scores. (2) quality assessments consistency, completeness, accuracy ten tables all exceed 0.80, variances between three dimensions being 0.04, 0.03, 0.05. overall evaluation items also exceeds consistent at table level. framework constructed this demonstrates high-quality stability. (3) Taking flood dataset as an example, clustering reveals five main themes high similarity within clusters, differences clusters are deemed significant relative significance level 0.01. Overall, experiment supports effective sharing resources enhances emergency response efficiency.

Язык: Английский

---