A file archival integrity check method based on the BiLSTM + CNN model and deep learning

Abstract

Validating and integrity-checking archives ensures that files are authentic, trustworthy, and usable. In the age of digital technology, historical records must be genuine. Researching in archives raises ethical issues while having little to do with individuals. Traditional archive integrity solutions have scaling issues, real-time monitoring issues, and missed opportunities. An updated Archive File Integrity Check Method (AFICM) may solve these issues, and the paper explains it. Deep learning allows the combination of a Bidirectional Long-Short Term Memory (Bi-LSTM) with adaptive gating and an adaptive Temporal Convolutional Neural Network (TCNN) with multi-scale temporal attention. This method protects archived material against manipulation, which is crucial. The recommended method extracts complex sequential patterns and variants using adaptive TCNN trained on file data. Next, it analyzes these features using a Bi-LSTM network and attenuation method. It allows it to highlight significant temporal correlations while downplaying irrelevant data selectively. The hybrid model outperforms checksums in accuracy and dependability. It uses adaptive TCNNs for time-related feature extraction and attenuated Bi-LSTM for refinement. The F1 score, recall, accuracy, precision, and AU-ROC are critical measures for model evaluation. The AICM performed well overall, with 97.32% precision and 98.95% accuracy. This integrity check method outperforms others with an F1 score of 97.58, an AU-ROC of 0.983, and a recall rate of 98.18%. The findings set a new standard for archiving system integrity testing by showing the model’s dependability and security in several use scenarios.