Advances in Nonlinear Variational Inequalities ISSN: 1092-910X Vol 28 No. 7s (2025) 664 https://internationalpubls.com Exploring Linguistic and Emotional Models for Audio Sentiment Analysis Using NLP 1Sapna Sharma, 2Ritu Sindhu Computer Science and Engineering Lingayas’s Vidhyapeeth Faridabad, India sharmagsapna@gmail.com Computer Science and Engineering Lingayas’s Vidhyapeeth Faridabad, India dr.ritusindhu@lingayasvidyapeeth.edu.in Article History: Received: 27-10-2024 Revised:11-11-2024 Accepted:19-12-2024 Abstract: Sentiment analysis is widely used to identify emotions and attitudes in text. With the growing popularity of audio-based social platforms and the significant rise in spoken data, sentiment analysis in the auditory domain has become increasingly important. This paper explores sentiment analysis in audio data using Natural Language Processing (NLP) techniques. We propose a novel method for extracting linguistic features and developing emotional models tailored to audio-based sentiment. In our experiments, we compare deep learning models with traditional NLP techniques, using a unique dataset to validate our findings. Sentiment analysis, also known as opinion mining, is a key subfield of NLP, focusing on extracting subjective information from textual data. The surge of user-generated content on online platforms like social media, blogs, and product reviews has amplified the importance of sentiment analysis for understanding public opinion and consumer behavior. This paper provides an overview of various approaches used in sentiment analysis, including machine learning, lexicon-based methods, and deep learning, highlighting their strengths and limitations. We discuss the trade-offs between accuracy, computational efficiency, and interpretability for each approach, while addressing challenges like sarcasm detection, context dependency, and domain-specific language. Additionally, we examine recent advancements in the field, such as the use of cross-sectional models and the integration of multiple data sources to provide a more comprehensive view of sentiment. Our results demonstrate the efficiency and high performance of the proposed models in capturing sentiment from audio data. The study also explores the ethical considerations, practical applications, and broader relevance of audio-based sentiment analysis across media and other domains. Finally, we conclude by discussing future directions, emphasizing the need for more robust models capable of handling diverse and complex data, along with ethical considerations for real-world applications. Keywords: Emotional Analysis, Natural Language Processing (NLP), Emotional Intelligence, Language Processing, Speech Analysis, Text-to-Speech (TTS), Speech Recognition, Image Extraction. Introduction The growing significance of deciphering human emotions and feelings through audio information is undeniable. With the rise of audible content on podcasts, audiobooks, social media platforms, and voice-activated devices, human emotions, attitudes, and opinions are increasingly being captured and shared through sound [1]. Sentiment analysis, the automated identification and classification of emotions or opinions in textual data [2], must now expand to accommodate the auditory realm [7]. Audio-based sentiment analysis introduces new layers of depth and nuance, capturing the subtleties of tone, emotion, and attitude that are often absent in text-based approaches [3]. However, processing, analyzing, and interpreting spoken language in audio data presents unique challenges, such as variations in speech patterns, accents, tone, and the presence of background noise [8]. To accurately identify and classify sentiments from audio signals, it is crucial to develop tailored methodologies and models. Advances in Nonlinear Variational Inequalities ISSN: 1092-910X Vol 28 No. 7s (2025) 665 https://internationalpubls.com This research seeks to create robust models capable of accurately and efficiently classifying sentiments in spoken language, thereby broadening the scope of sentiment analysis to encompass audio data. By leveraging NLP techniques specifically adapted for audio analysis, this study aims to bridge the gap between audio data and sentiment analysis, offering deeper insights into human emotions conveyed through sound [4]. The applications of this research are vast, including customer feedback analysis, emotional well- being assessments, and sentiment-aware user interfaces, among others [11]. Ultimately, our goal is to develop machines that can understand and respond to human emotions, enhancing human-computer interactions and fostering a more empathetic technological environment [2]. Methodology: 1. Data Collection and Preprocessing: • Audio Data Acquisition: Gather a large set of audio clips from various sources such as podcasts, interviews, customer service calls, and social media snippets. To build a robust model, it is essential to ensure diversity in this collection, including a range of emotions, languages, accents, and contexts. • Audio-to-Text Conversion: Use Automatic Speech Recognition (ASR) systems to convert the audio data into text, enabling further analysis through NLP techniques. • Data Cleaning and Preprocessing: Clean the data by removing unnecessary elements such as extra characters and special symbols, and normalize capitalization. For audio data, remove background noise, segment the audio into smaller units, and extract key audio features. 2. Feature Extraction: • Lexical Features: Extract specific text features, such as n-grams, TF-IDF vectors, and word embedding’s (e.g., Word2Vec or GloVe). Pre-trained language models (like BERT or GPT) can also be employed to enhance the contextual meaning of the text. • Audio Features: Extract key audio features, including pitch, Mel-Frequency Campestral Coefficients (MFCC), and prosodic features. These audio features are then combined with text features to create a hybrid representation that captures both the linguistic and auditory elements of the data. 3. Model Development and Training: • Model Selection: Choose appropriate models for analysis, such as Recurrent Neural Networks (RNN), Convolutional Neural Networks (CNN), Long Short-Term Memory networks (LSTM), or transformers. The structural design of the models should be adapted to analyze both textual and auditory data. • Training and Evaluation: Split the preprocessed data into training, validation, and test sets to train and test the neural network models. Evaluate the model's performance using metrics like accuracy, precision, recall, F1-score, and confusion matrix. 4.Testing and Optimization: • Parameter Tuning: Optimize model parameters such as learning rate, batch size, and model architecture using techniques like grid search or random search to fine-tune performance. • Comparative Analysis: Compare the performance of the proposed model with other baseline models that do not incorporate audio features. Highlight the advantages and unique contributions of the proposed approach to emotion analysis in the audio domain. Advances in Nonlinear Variational Inequalities ISSN: 1092-910X Vol 28 No. 7s (2025) 666 https://internationalpubls.com Fig1: Process of sentiment analysis LITERATURE REVIEW Machine learning (ML) and natural language processing (NLP) applications for analyzing unstructured patient feedback have garnered significant attention due to the valuable insights they provide. This systematic review covers 19 articles, revealing that 80% of the studies focus primarily on the linguistic analysis of unsolicited patient comments from social media platforms, with formal surveys (solicited feedback) being the second most common source. Supervised learning was the most frequently used method (n = 9), followed by unsupervised (n = 6) and semi-supervised (n = 3) approaches. Social media comments were typically analyzed using unsupervised methods, while free-text responses from structured questionnaires were processed using supervised approaches [11]. The most effective machine learning classifiers were found to be Support Vector Machines and Naïve Bayes, based on performance metrics such as precision, recall, and F-measure. These findings emphasize the growing importance of NLP and ML as essential tools for managing unstructured patient feedback. The choice between supervised and unsupervised approaches depends largely on the data source, reflecting the diversity in patient feedback channels. As healthcare organizations face increasing volumes of unstructured text data, advanced data analysis tools present promising opportunities to extract meaningful insights. Ongoing advancements in these tools are expected to further enhance the utility of NLP and ML techniques, providing healthcare organizations with robust methods for analyzing patient feedback. Sarcopenia, characterized by age-related muscle loss, can be addressed effectively through physical activity. However, adherence to home-based exercise programs is low, at around 40%, necessitating the development of solutions to monitor progress and boost compliance. The integration of remote sensing systems in mobile applications, designed with user input, shows potential, particularly when combined with NLP for data collection. In a study involving 22 participants (average age 68), feedback from both patients and physicians informed the design of the app. System usability (SUS) scores averaged 66.4 (SD 13.6), while user satisfaction with the technology (USE) averaged 41.3 (SD 16.2). Positive emotions during interviews correlated with higher SUS scores (β = 1.45; 96% confidence interval (CI): 0.42 to 2.45; P = .01), underscoring the impact of user experience on technology acceptance. However, emotions did not significantly affect USE scores. Latent Dirichlet Analysis (LDA) revealed distinct themes between patients and physicians, indicating differences in their perspectives on the program's needs. Contact centers play a vital role in organizations, and the COVID-19 pandemic has highlighted their importance in maintaining business continuity, financial performance, and customer support. The surge in pandemic-related calls has prompted organizations to reevaluate call center operations. Many are now adopting next-generation machine learning and NLP platforms, including self-service portals and chatbots, to enhance customer service. This literature review explores the shift toward Advances in Nonlinear Variational Inequalities ISSN: 1092-910X Vol 28 No. 7s (2025) 667 https://internationalpubls.com these innovative solutions, aiming to identify research gaps, highlight the benefits of NLP integration, and address the challenges faced by contact center organizations. By examining the advantages and obstacles, this review provides recommendations for accelerating contact center automation and improving customer support technologies. In recent years, NLP has emerged as a cornerstone of artificial intelligence, driving significant advancements in machine learning and language processing. One of the key applications of NLP is machine translation, which plays a vital role in human-computer interaction. The field of NLP has progressed substantially, especially with the development of neural network architectures that provide a strong mathematical and theoretical foundation. Sequential analysis in language processing has paved the way for the exploration of neural networks and established a framework for multilingual processing, which is particularly relevant in multilingual contexts. Linguistic Convolutional Neural Networks (CNNs) have been developed to cater to speakers of different languages, aiming to provide more accessible information for native speakers. This study examines the impact of recent advancements in English language processing on language development. Using computational methods, it proposes a model to simplify the conversion of spoken language into written text through probabilistic models that account for the complexity of language structure. This approach addresses the challenges posed by differences in body language and contributes to broader language transformations. Recent research has focused on implementing CNNs and attention models, which have proven effective in cross-language tasks. The statistical distribution of these models and their performance in various linguistic contexts are discussed. This paper also highlights the challenges of achieving accurate speech recognition in noisy environments, emphasizing the difficulty of eliminating interference in language processing. The integration of open learning and text mining for the automated analysis of customer feedback has gained attention in recent research. Language-based text mining models, such as those used in the ARC framework, have shown effectiveness in capturing language patterns, with a reported accuracy of 92%. This highlights the utility of text mining in understanding customer opinions and identifying key areas in service processes that generate compliments and complaints. For organizations, the speed and accuracy of these models are crucial for quickly diagnosing and addressing customer concerns, providing insights that go beyond traditional quantitative research. Our primary research objective is to develop a robust and accurate emotion analysis model for audio data using NLP techniques. This model will process and analyze audio signals, extracting useful features for emotion classification to enhance the understanding of human speech. Getting this process right is key to unlocking the full potential of emotion analysis in the audio domain, improving applications that rely on such analysis. Limitations: 1. Data Quality Audio data and emotional signals can vary in quality, which can pose challenges for our models to learn effectively. Ensuring the accuracy of data processing may require additional time and effort to refine the data before progressing further. 2. Model Development Our model may struggle to handle all types of data and audio languages effectively. Adapting it to speech-to-text NLP techniques is proving to be more complex than initially expected. This complexity could slow our progress and may limit the scalability of the model. Advances in Nonlinear Variational Inequalities ISSN: 1092-910X Vol 28 No. 7s (2025) 668 https://internationalpubls.com 3. Resource Constraints The available computational power and time may not be sufficient to accomplish all objectives in this study. These limitations could restrict the size of the tests we can conduct or reduce the complexity of the models we are able to test. 4. Evaluation Metrics The standard evaluation metrics we currently use may not capture all the nuances of audio analysis. This makes it difficult to fully understand how well our model performs. We may need to explore alternative metrics or simpler ways to evaluate model performance. Three-Class Sentiment Analysis To handle three-class sentiment analysis (positive, negative, neutral), you can follow the steps below: 1. Problem Definition Objective: Classify text into one of three categories—positive, negative, or neutral sentiment. 2. Data Collection • Datasets: You can use publicly available datasets or collect your own. Examples include: • Sentiment140: Contains tweets labeled as positive, negative, or neutral. • Yelp Reviews: Reviews labeled as positive, negative, or neutral. • Custom Datasets: You can create your own dataset by labeling specific text data. 3. Data Preprocessing • Remove URLs, usernames (e.g., @username), and hashtags. • Remove punctuation, numbers, and special characters. • Convert all text to lowercase for uniformity. • Tokenize the text into words or symbols. • Optional: • Stop word removal: Remove common words that don’t contribute much to sentiment (e.g., "the," "is," "and"). • Lemmatization/Stemming: Reduce words to their base form (e.g., "running" to "run"). 4. Feature Extraction • Bag of Words (BoW): Convert text into numeric representations. • TF-IDF (Term Frequency-Inverse Document Frequency): Measure the importance of terms within the dataset. • Word Embeddings: Use pre-trained embeddings like Word2Vec, GloVe, or BERT for better context understanding. 5. Model Selection • Logistic Regression: Suitable for simple and interpretable models. • Naive Bayes: Effective for text classification. • Support Vector Machines (SVM): Performs well in high-dimensional spaces. Advances in Nonlinear Variational Inequalities ISSN: 1092-910X Vol 28 No. 7s (2025) 669 https://internationalpubls.com • Deep Learning Models: • LSTM/GRU: Useful for sequential data, especially when context matters. • CNN: Good for capturing local (n-gram) patterns. • BERT/Transformers: Advanced models capable of understanding context and relationships between words. 6. Training the Model • Data Split: Divide your dataset into training, validation, and test sets (e.g., 80% training, 10% validation, 10% testing). • Training: Train the selected model on the training set. • Hyperparameter Tuning: Use the validation set to fine-tune model parameters (e.g., learning rate, batch size). 7. Model Evaluation • Metrics: Evaluate model performance using metrics like precision, accuracy, recall, F1-score, and confusion matrix. • Confusion Matrix: Track true positives, true negatives, false positives, and false negatives for each sentiment class (positive, negative, neutral). • Cross-Validation: Use k-fold cross-validation to optimize model performance on the dataset. 8. Application • Model Deployment: Use the trained model to classify new text data over time. • API Integration: Develop an API that allows other applications to leverage the model for sentiment analysis. 9. Model Enhancement • Data Augmentation: Increase the dataset size using techniques like word substitution, back- translation, or paraphrasing. • Ensemble Models: Combine predictions from multiple models to improve overall performance. • Fine-Tuning: Refine pre-trained models (e.g., BERT) on your specific dataset to boost accuracy. Fig 2. Normalizes matrix of three class sentiment analysis. Advances in Nonlinear Variational Inequalities ISSN: 1092-910X Vol 28 No. 7s (2025) 670 https://internationalpubls.com Comparison of various approaches: 1. LSTM: Effective for capturing complex patterns and sequences in large datasets. 2. Naive Bayes: Simple and fast, making it an ideal starting point for smaller datasets. 3. SVM: Highly effective for high-dimensional data, especially when using nonlinear models. 4. Logistic Regression: A reliable model that offers fast training and easy interpretability. 5. Decision Tree: Useful for data profiling, but may require advanced techniques to achieve competitive performance. Model Strengths Weaknesses Best Used For Performance LSTM It records sequence, context and long-term dependencies. It is complex, resource-intensive, and requires large datasets. Sequential data, complex context Have large datasets and complex samples. Naive Bayes It's simple, fast, and works well on small datasets. With autonomy, there is little understanding of context. The base, smaller data set, is spatial analysis. It is correct, but works better on larger datasets. SVM It handles high- dimensional data well and is generalizable. Computer literate and need to be well organized. Higher resolution limits and nonlinearities. High and fair work ethic. Logistic Regression Simple, clear and effective. As a linear hypothesis, there is little functional interaction. Line segmentation data, baseline. It is robust, especially in large and flexible datasets. Decision Tree The translation captures non- linear relationships. Excess, volatility, and bias in unbalanced data. Translation, non- linear relationships. On average, it often improves team culture. Table no 1. Comparision of various approaches The matrix of comparison show in graph Fig 3. Matrix of comparison of various approaches. The accuracy of sentiment analysis relies on data quality, size, classifier type, and preprocessing techniques, ultimately affecting classifier performance. Advances in Nonlinear Variational Inequalities ISSN: 1092-910X Vol 28 No. 7s (2025) 671 https://internationalpubls.com Classifier Dataset Accuracy Model 1 Amazon-Fine-Food Reviews 87 Cell Phones and Accessories 88 Amazon Products 98 IMDB 85 Model 2 Amazon-Fine-Food Reviews 96 Cell Phones & Gadgets 87 Amazon Products 95 IMDB 79 Model 3 Amazon-Fine-Food Reviews 89 Cell Phones and Accessories 87 Amazon Products 97.5 IMDB 78 Fig 4. Accuracy of different models. Future Research Directions: 1. Multimodal Analysis: Investigate methods to integrate audio, text, and video for a comprehensive understanding of emotions. This approach can help simultaneously capture facial expressions, gestures, and vocal cues. 2. Deep Learning Architectures: Explore advanced deep learning architectures to analyze complex and subtle audio patterns. Transfer learning techniques could also be leveraged to enhance model performance. 3. Unsupervised and Supervised Learning: Develop strategies to reduce dependence on large datasets and extend sentiment analysis to low-resource languages and regions. 4. Domain-Specific Sentiment Analysis: Build sentiment analysis models tailored for specific domains, such as healthcare, finance, or education, to accurately capture region or industry-specific emotions. Advances in Nonlinear Variational Inequalities ISSN: 1092-910X Vol 28 No. 7s (2025) 672 https://internationalpubls.com 5. Real-Time Analytics: Investigate techniques for real-time or streaming sentiment analysis to analyze emotions in live audio streams or ongoing conversations. 6. Deeper Understanding of Human Emotions: Delve into more profound emotional understanding by integrating sentiment analysis into broader emotional frameworks. 7. Transfer Learning Across Languages: Explore transfer learning methods to apply sentiment analysis knowledge from one language to another, particularly for underrepresented or minority languages. 8. Ethical Considerations: Focus on identifying and mitigating biases in sentiment analysis models, particularly regarding gender, race, and culture, to ensure fair and accurate emotion classification. 9. User-Centered Applications: Research user-centered analytical tools, such as recommendation systems and intelligent user interfaces, to optimize products and services for enhanced user experience. Conclusion: Our research addresses a crucial gap in sentiment analysis by developing a novel model tailored for detecting anomalies in audio data using advanced natural language processing (NLP) techniques. The primary objective is to bridge the gap between text-based emotion analysis and the growing need for evaluating emotions in the fast-evolving domain of audio data. Given the unique challenges posed by audio, such as variations in speech intonation, tone, accents, and background noise, a specialized approach to sentiment analysis is essential. Through the adaptation of NLP techniques, we have refined existing models to efficiently process audio data and extract key features for sentiment classification. Our experimental results demonstrated the effectiveness of our model in accurately classifying emotions across various audio datasets. We observed significant improvements in audio sentiment analysis, confirming the validity of our proposed approach. Notably, the model performs exceptionally well across multiple languages and audio formats, demonstrating its adaptability and versatility. However, the study also highlights certain limitations, particularly the lack of available audio datasets in specific languages and regions, which may hinder progress. Additionally, optimizing the model for peak performance requires a deeper understanding of the acoustic and emotional properties involved, encouraging further exploration and experimentation. The implications of this research are extensive. Accurate sentiment analysis in audio data has the potential to unlock new insights across numerous applications, including customer feedback analysis, mental health monitoring, and the development of responsive user interfaces. The ability to accurately interpret emotions and attitudes conveyed through speech has the potential to transform how we interact with technology and how we understand human behavior. References [1] Sen, A., Rajakumaran, G., Mahdal, M., Usharani, S., Rajasekharan, V., Vincent, R., & Sugavanan, K. (2024). Live Event Detection for People’s Safety using NLP and Deep Learning. IEEE Access. [2] Hassan, N., Miah, A. S. M., & Shin, J. (2024). A Deep Bidirectional LSTM Model Enhanced by Transfer-Learning- Based Feature Extraction for Dynamic Human Activity Recognition. Applied Sciences, 14(2), 603. [3] Kumar, L. K., Thatha, V. N., Udayaraju, P., Siri, D., Kiran, G. U., Jagadesh, B. N., & Vatambeti, R. (2024). Analyzing Public Sentiment on the Amazon Website: A GSK-based Double Path Transformer Network Approach for Sentiment Analysis. IEEE Access. [4] Sharma, P., & Dash, B. (2023). Active Noise Cancellation in Microsoft Teams Using AI & NLP Powered Algorithms. AIRCC's International Journal of Computer Science and Information Technology, 31-42. Advances in Nonlinear Variational Inequalities ISSN: 1092-910X Vol 28 No. 7s (2025) 673 https://internationalpubls.com [5] Bowler, L., Lopatovska, I., & Rosin, M. S. (2023). Teen-adult interactions during the co-design of data literacy activities for the public library: insights from a natural language processing analysis of linguistic patterns. Information and Learning Sciences. [6] Sarmet, M., Kabani, A., Coelho, L., Dos Reis, S. S., Zeredo, J. L., & Mehta, A. K. (2023). The use of natural language processing in palliative care research: A scoping review. Palliative Medicine, 37(2), 275-290. [7] Kumar, M., Janani, P., Kumaragurubaran, K., & Sundari, P. (2023). Automated Notes Maker from Audio Recordings. Journal of Image Processing and Intelligent Remote Sensing (JIPIRS) ISSN 2815-0953, 3(03), 22-26. [8] Joel, M. R., Ebenezer, V., Jeyaraj, K. A., Navaneethakrishnan, M., Arunadevi, R., & Jenifa, D. R. (2023, February). Decoding and Analysing Consumer Feedback for Companies and Goods using Machine Learning. In 2023 Third International Conference on Artificial Intelligence and Smart Energy (ICAIS) (pp. 1085-1091). IEEE. [9] Shah, S., Ghomeshi, H., Vakaj, E., Cooper, E., & Fouad, S. (2023). A review of natural language processing in contact centre automation. Pattern Analysis and Applications, 26(3), 823-846. [10] Khan, M. U., Javed, A. R., Ihsan, M., & Tariq, U. (2023). A novel category detection of social media reviews in the restaurant industry. Multimedia Systems, 29(3), 1825-1838. [11] Kumar, L. K., Thatha, V. N., Udayaraju, P., Siri, D., Kiran, G. U., Jagadesh, B. N., & Vatambeti, R. (2024). Analyzing Public Sentiment on the Amazon Website: A GSK-based Double Path Transformer Network Approach for Sentiment Analysis. IEEE Access. [12] Symum, M. A. A., Sheemu, S. Y., Asif, A. S. M., & Islam, K. (2023). Sentimental analysis of customer product Reviews to understand customer needs using machine learning (Doctoral dissertation, Brac University). [13] Iqbal, A., Amin, R., Iqbal, J., Alroobaea, R., Binmahfoudh, A., & Hussain, M. (2022). Sentiment analysis of consumer reviews using deep learning. Sustainability, 14(17), 10844. [14] Ahmed, A. A. A., Agarwal, S., Kurniawan, I. G. A., Anantadjaya, S. P., & Krishnan, C. (2022). Business boosting through sentiment analysis using Artificial Intelligence approach. International Journal of System Assurance Engineering and Management, 13(Suppl 1), 699-709. [15] Trisna, K. W., & Jie, H. J. (2022). Deep learning approach for aspect-based sentiment classification: a comparative review. Applied Artificial Intelligence, 36(1), 2014186. [16] Jain, P. K., Pamula, R., & Srivastava, G. (2021). A systematic literature review on machine learning applications for consumer sentiment analysis using online reviews. Computer science review, 41, 100413. [17] Mendi, A. F. (2022). A sentiment analysis method based on a blockchain-supported long short-term memory deep network. Sensors, 22(12), 4419. [18] Dang, C. N., Moreno-García, M. N., & Prieta, F. D. L. (2021). An approach to integrating sentiment analysis into recommender systems. Sensors, 21(16), 5666. [19] Hawladar, M., Ghosh, A., Raad, Z. K., Chowdhury, W. A., & Shehan, M. S. H. (2021). Amazon product reviews sentiment analysis using supervised learning algorithms (Doctoral dissertation, Brac University). [20] Jain, P. K., Pamula, R., & Srivastava, G. (2021). A systematic literature review on machine learning applications for consumer sentiment analysis using online reviews. Computer science review, 41, 100413. [21] Chatterjee, I. (2021). Statistics-Based Anomaly Detection and Correction Method for Amazon Customer Reviews (Doctoral dissertation, New Jersey Institute of Technology). [22] Birjali, M., Kasri, M., & Beni-Hssane, A. (2021). A comprehensive survey on sentiment analysis: Approaches, challenges and trends. Knowledge-Based Systems, 226, 107134. [23] Petersen, C. L., Halter, R., Kotz, D., Loeb, L., Cook, S., Pidgeon, D., ... & Batsis, J. A. (2020). Using natural language processing and sentiment analysis to augment traditional user-centered design: development and usability study. JMIR mHealth and uHealth, 8(8), e16862. [24] Kumar, A., & Jaiswal, A. (2020). Systematic literature review of sentiment analysis on Twitter using soft computing techniques. Concurrency and Computation: Practice and Experience, 32(1), e5107. [25] Jain, P. K., Pamula, R., & Srivastava, G. (2021). A systematic literature review on machine learning applications for consumer sentiment analysis using online reviews. Computer science review, 41, 100413. [26] Thaler, F., Faußer, S., & Gewald, H. (2021). Using NLP to analyze whether customer statements comply with their inner belief. arXiv preprint arXiv:2107.11175. Advances in Nonlinear Variational Inequalities ISSN: 1092-910X Vol 28 No. 7s (2025) 674 https://internationalpubls.com [27] Shoumy, N. J., Ang, L. M., Seng, K. P., Rahaman, D. M., & Zia, T. (2020). Multimodal big data affective analytics: A comprehensive survey using text, audio, visual and physiological signals. Journal of Network and Computer Applications, 149, 102447. [28] Aljuhani, S.A.; Alghamdi, N.S. A Comparison of Sentiment Analysis Methods on Amazon Reviews of Mobile Phones. Int. J. Adv. Comput. Sci. Appl. 2019, 10, 608–617. [29] Shoumy, N. J., Ang, L. M., Seng, K. P., Rahaman, D. M., & Zia, T. (2020). Multimodal big data affective analytics: A comprehensive survey using text, audio, visual and physiological signals. Journal of Network and Computer Applications, 149, 102447. [30] Nguyen, H.; Veluchamy, A.; Diop, M.; Iqbal, R. Comparative Study of Sentiment Analysis with Product Reviews Using Machine Learning and Lexicon-Based Approaches. SMU Data Sci. Rev. 2018, 1, 7. [31] Seo, S., Kim, C., Kim, H., Mo, K., & Kang, P. (2020). Comparative study of deep learning-based sentiment classification. IEEE Access, 8, 6861-6875.