An Enhanced and High-Performance Algorithm for Accurate Named Entity Recognition in Legal Text

The rise of digital technologies results in the generation of large volumes of legal data. Although natural language legal texts are easily understood by humans, they pose significant challenges for machine processing. Converting this information into a machine-readable format is essential for efficient legal data analysis. In this paper, we propose resource-efficient method for extracting and classifying named entities within legal documents. We use rule-based learning and regular expressions to identify legal entities like parties, courts, and case references. Once extracted, these entities are classified into categories such as individuals, legal organizations, and jurisdictions. To improve both accuracy and processing speed, we implement a modified K-Means algorithm clustering the extracted entities. This clustering step refines the classification by grouping entities based on their shared attributes. The proposed method significantly enhances the speed and precision of Named Entity Recognition (NER) for legal texts, enabling faster and more efficient legal data processing.