Enhancing AI Classification Performance: A Framework for Textual Data Quality Assessment

The rapid advancement of AI technologies has led to increased reliance on classification-based systems across various domains. However, the quality of textual data used in these systems significantly impacts their performance, reliability, and fairness. While quality standards exist for general data and AI systems, a comprehensive framework specifically addressing textual data quality in classification tasks is lacking. This study aims to develop a robust framework for assessing and improving textual data quality in AI-based classification systems. The research employs a mixed-methods approach, including literature review, expert consultations, and empirical validation. The proposed framework encompasses key quality dimensions, metrics, and improvement methodologies tailored to textual data challenges. By addressing the critical role of data quality in AI system development, this study seeks to enhance the reliability, fairness, and overall performance of classification-based AI systems. The framework’s validation using real-world datasets demonstrates its practical applicability and potential impact on AI model outcomes.