PII Data Detection using Keras & NLP

The research presents a reliable and effective method for automatically identifying and categorizing PII items in various textual datasets using advanced deep learning models, specifically KerasNLP, DeBERTaV3, and DistilBERT. Sensitive information including names, addresses, and social security numbers are included in personally identifiable information. The research makes use of neural network technology to automatically recognize and categorize PII elements found in textual data. A comprehensive method that includes training strategies specifically designed for PII detection, model architecture design, and preprocessing stages. The proposed model has demonstrated its ability to accurately identify Personally Identifiable Information (PII) by achieving an accuracy of 89.8% and a minimal loss of 0.0004% when adjusting to longer input sequences.