The widespread use of online social media has enabled users to express their thoughts, feelings, opinions, and sentiments in their preferred languages. These diverse perspectives offer valuable insights for data-driven decision-making. While extensive sentiment analysis approaches have been developed for resource-rich languages like English and Chinese, low-resource languages such as Roman Urdu and Roman Punjabi, especially in code-mixed contexts, have been largely neglected due to the lack of datasets and limited research on their unique morphological structures and grammatical complexities. This study aims to present a novel approach for multiclass sentiment analysis of low-resource, code-mixed datasets using multilingual transformers. Specifically, a dataset comprising Roman Urdu, Roman Punjabi, and English comments was collected. After applying traditional natural language preprocessing techniques, transformer-based libraries were used for tokenization and embedding. Subsequently, the Multilingual Bidirectional Encoder Representations from Transformers (mBERT) model was optimized and trained for multiclass sentiment analysis on the code-mixed data. The evaluation results showed a significant improvement in accuracy (+22.55%), precision (+21.06%), recall (+22.55%), and F-measure (+25.50%) compared to benchmark algorithms. Additionally, the proposed model outperformed other transformer-based models, as well as deep learning and machine learning algorithms in sentiment extraction from code-mixed data. These findings highlight the potential of the proposed approach for sentiment analysis in low-resource, code-mixed languages.

With the rapid proliferation of user-generated content across internet-based platforms, sentiment analysis has become an essential tool for understanding public opinion. However, sentiment analysis on code-mixed, low-resource languages presents significant challenges due to limited annotated data, linguistic complexity, and resource constraints. This systematic literature review (SLR) comprehensively examines recent developments in sentiment analysis for code-mixed and low-resource languages, with a focus on studies published between 2017 and 2025. The review investigates a wide range of approaches, including deep learning models, transfer learning strategies, and pre-trained language models, evaluating their effectiveness in low-resource and code-mixed contexts. Findings indicate that transfer learning and transformer-based architectures are increasingly preferred due to their superior performance and reduced reliance on large annotated datasets. Social media is identified as the dominant data source, followed by reviews from e-commerce and entertainment domains. Despite these advancements, significant challenges persist, including the scarcity of high-quality labeled data, high computational costs, and limited generalization across languages and domains. This review not only synthesizes current methodologies and applications but also proposes a conceptual framework for future research. It offers actionable insights for scholars seeking to develop efficient, scalable sentiment analysis systems tailored to code-mixed and low-resource language scenarios.