Fake Job Posting Detection using AI

The explosion of online job platforms has, unfortunately, opened the floodgates for fake job
ads. For job seekers, this isn’t just an annoying hassle—it’s a real threat, leading to scams, stolen
identities, and financial losses. Our research takes this problem head-on. We use advanced Machine
Learning (ML) and Natural Language Processing (NLP) tools to pull apart nearly 18,000 job ads from
the EMSCAD dataset. A big challenge is that the overwhelming majority of postings are real—about
95%—so we tackle this imbalance using SMOTE, which creates synthetic samples of rare fraudulent ads.
Then, with TF-IDF bigram features, we turn text into data for a Random Forest model. The results speak
for themselves: 96% accuracy with 97% recall when spotting scams.