SACM - United States of America
Permanent URI for this collectionhttps://drepo.sdl.edu.sa/handle/20.500.14154/9668
Browse
3 results
Search Results
Item Restricted Leveraging LLMs for the Analysis of Mobile App User Feedback: In-Depth Evaluation of User Perspectives on AI-Enabled Mobile Apps(Saudi Digital Library, 2025-05-10) Alsanousi, Bassam Jameel A; Ludi, Stephanie; Do, HyunsookThe expanding use of artificial intelligence (AI) in mobile applications has intensified the need to investigate how integrated AI features impact user experience (UX). While research in this area is growing, a significant gap exists in evaluating the usability of AI-enabled apps across languages, platforms, and domains. Furthermore, analyzing large-scale user feedback remains challenging despite the automation potential of recent large language models (LLMs). Interestingly, evaluating mobile apps using International Organization for Standardization (ISO) standards in exploring UX has shown promise while uncovering weaknesses and emerging issues. Accordingly, in this study, we evaluated mobile apps by analyzing user reviews through the lens of the ISO 9241-11 usability model and the ISO/International Electrotechnical Commission’s (IEC’s) 25010 quality standard. This dissertation has two main objectives. The first is to examine the performance of AI-enabled apps across different domains and platforms (iOS and Android) in multiple languages to identify emerging usability issues. The second objective is to develop trustworthy automated tools using LLMs and ISO standards that improve the semantic analysis of user feedback regarding usability and software quality and thus support the handling of large amounts of data. Our research results provide valuable insights into the benefits and difficulties of AI-enabled mobile apps in various domains. By conducting sentiment analysis, we find that users are generally positive about these apps; however, there are critical issues underlying the negative reviews related to AI (e.g., giving unclear responses, algorithmic bias, privacy concerns, voice and image recognition limitations, ethical sensitivity, and insufficient transparency in AI decision-making processes). Furthermore, the advanced tools we developed demonstrate their effectiveness in automatically analyzing user reviews according to the ISO standards compared to other advanced models (e.g., GPT-4o, Llama2, and Gemini). In addition, our research has uniquely applied interpretability techniques—local interpretable model-agnostic explanations (LIME)—to develop LLMs capable of interpreting their output, aiding in the creation of trustworthy models. These findings provide developers, app owners, and researchers with insights into user perceptions of AI-enabled apps while presenting advanced strategies for automating the analysis of user reviews effectively.19 0Item Restricted AUTOMATED DETECTION OF OFFENSIVE TEXTS BASED ON ENSEMBLE LEARNING AND HYBRID DEEP LEARNING TECHNIQUES(Florida Atlantic University, 2025-05) Alqahtani, Abdulkarim Faraj; Ilyas, MohammadThe impact of communication through social media is currently considered a significant social issue. This issue can lead to inappropriate behavior using social media, which is referred to as cyberbullying. The accessibility and freedom of expression afforded by social media platforms enable individuals to share their emotions and opinions, but it also leads to cyberbullying behavior. Automated systems are capable of efficiently identifying cyberbullying and performing sentiment analysis on social media platforms. In this dissertation, our focus is on enhancing a system to detect cyberbullying in various ways. Therefore, we apply natural language processing techniques utilizing artificial intelligence algorithms to identify offensive texts in various public datasets. The first approach leverages two deep learning models to classify a large-scale dataset, combining two techniques: data augmentation and the GloVe pre-trained word representation method to improve training performance. In addition, we utilized multi-classification algorithms on a cyberbullying dataset to identify six types of cyberbullying tweets. Our approach achieved high accuracy, particularly with TF-IDF (bigram) feature extraction, compared to previous experiments and traditional machine learning algorithms applied to the same dataset. We employed two ensemble machine learning methods with the TF-IDF feature extraction technique, which demonstrated superior classification performance. Moreover, we used four feature extraction methods, BoW, TF-IDF, Word2Vec, and GloVe, to determine which works best with the ensemble technique. Finally, we utilize a multi-channel convolutional neural network (CNN) enhanced with an attention mechanism and optimized using a focal loss function.16 0Item Restricted IMPROVING ASPECT-BASED SENTIMENT ANALYSIS THROUGH LARGE LANGUAGE MODELS(Florida state university, 2024) Alanazi, Sami; Liu, XiuwenAspect-Based Sentiment Analysis (ABSA) is a crucial task in Natural Language Processing (NLP) that seeks to extract sentiments associated with specific aspects within text data. While traditional sentiment analysis offers a broad view, ABSA provides a fine-grained approach by identifying sentiments tied to particular aspects, enabling deeper insights into user opinions across diverse domains. Despite improvements in NLP, accurately capturing aspect-specific sentiments, especially in complex and multi-aspect sentences, remains challenging due to the nuanced dependencies and variations in sentiment expression. Additionally, languages with limited annotated datasets, such as Arabic, present further obstacles in ABSA. This dissertation addresses these challenges by proposing methodologies that enhance ABSA capabilities through large language models and transformer architectures. Three primary approaches are developed and evaluated: First, aspect-specific sentiment classification using GPT-4 with prompt engineering to improve few-shot learning and in-context classification; second, triplet extraction utilizing an encoder-decoder framework based on the T5 model, designed to capture aspect-opinion-sentiment associations effectively; and lastly, Aspect-Aware Conditional BERT, an extension of AraBERT, incorporating a customized attention mechanism to dynamically adjust focus based on target aspects, particularly improving ABSA in multi-aspect Arabic text. Our experimental results demonstrate that these proposed methods outperform current baselines across multiple datasets, particularly in improving sentiment accuracy and aspect relevance. This research contributes new model architectures and techniques that enhance ABSA for high-resource and low-resource languages, offering a scalable solution adaptable to various domains.38 0