Automated Detection of Fraudulent Job postings and Resumes Using NLP and ML Techniques

The project focuses on developing an automated system for detecting fraudulent job postings and classifying resumes using machine learning and natural language processing techniques. Fraudulent job postings often contain misleading information that can affect recruitment decisions, while unstructured resumes make it difficult to organize and analyze candidate data efficiently. This system integrates multiple algorithms to address these challenges. For job posting fraud detection, Random Forest, XGBoost, and Convolutional Neural Networks are employed to analyze textual and structured data, extracting relevant features and identifying patterns associated with fraudulent content. For resume classification, natural language processing techniques combined with BERT (Bidirectional Encoder Representations from Transformers) are used to understand context and categorize resumes accurately. The system is implemented as a web-based platform using the Flask framework, providing modules for user registration, login, job fraud detection, resume classification, and secure logout. Text preprocessing, feature extraction, model training, and evaluation are key steps in the pipeline, ensuring reliable and precise predictions. Experimental results demonstrate the effectiveness of ensemble learning and transformer-based models in handling textual data for classification and fraud detection tasks. The system provides an organized interface for users to submit job postings or resumes and receive predictions in accurate. This approach can be extended to other domains involving textual document verification and categorization, making it a scalable and adaptable solution for structured data management. Overall, the project combines machine learning and NLP techniques to improve data validation, classification accuracy, and system usability.

Keywords: Fraud detection, Job postings, Resume classification, Machine learning, NLP, Random Forest, XGBoost, CNN, BERT, Flask.