An english financial audit anomaly detection system for construction industry based on machine learning

The complexity of project-based financial flows, the use of English voluminous textual records, and invoice, contract, and progress report inconsistencies make financial auditing in the construction industry a challenging process. However, existing machine learning-based auditing approaches still suffer from limitations such as high false-positive rates, weak adaptability to dynamic financial data, and insufficient integration of statistical and semantic textual features. This research aims to develop a Machine Learning (ML)-Based Anomaly Detection System for English financial auditing in construction projects using advanced text processing, hybrid feature extraction, and an optimized one-class classification model to improve anomaly identification accuracy. Financial documents, including statements, invoices, and transaction records, were collected, totaling 2,400 construction financial records. Data preprocessing involved text normalization and handling missing values to ensure data quality. Feature extraction employed TF–IDF vectors for statistical term importance and Word2Vec embeddings for semantic relationships. To address the identified research gap, a hybrid Big Bang–Big Crunch Optimized Adaptive Weighted One-Class Support Vector Machine (3BCO-AW-OCSVM) model was proposed to enhance classification boundary optimization and improve sensitivity toward rare financial irregularities. 3BCO optimized model parameters through expansion and contraction, while AW-OCSVM adapted to normal financial behavior with weighted emphasis on informative samples. The consistency and reliability of the model over validation folds were statistically tested in paired t-tests. Experimental results demonstrate superior performance, with anomaly detection metrics ranging from 0.95 to 0.97, outperforming conventional ML approaches. The proposed system improved anomaly detection reliability, reduced false positives, and supported scalable real-time financial auditing in construction projects.