Development of a Fake Job Posting Detection System using Deep Neural Networks and Voting Ensemble Methods

Abstract

The increase in fake job postings online, especially in places such as Nigeria, is a big problem for people searching for jobs. To address this, we designed and built a Deep Neural Network (DNN) ensemble model to find these deceptive job ads. We used a dataset from Kaggle that had both real and fake job listings. The data was cleaned by removing unnecessary information, changing all text to lowercase, and getting rid of common words. The DNN was then trained to tell the difference between real and fake postings. It did well, with 96% accuracy, 94% precision, 96% recall, and an AUC of 98%. These result shows that the model is good at picking out fake listings from real ones. To make it easy to use, the model was put into a web application. This application was created using the Waterfall Model to keep things organized. The app lets people find fake postings in real time, which helps them stay safe and trust the platform. There is also a dashboard for administrators to watch over and manage job postings, making the system easy to use and dependable. This Paper shows that combining machine learning with web solutions can solve actual problems and make online job markets safer and more trustworthy.