Have you ever found yourself drowning in a sea of survey responses, wondering which ones are genuine and which are generated by bots? I recently stumbled upon a Reddit post from someone who’s facing this exact problem. They’ve been tasked with identifying bot-generated responses in a large dataset, but they’re not sure where to start.
Their initial research points them towards anomaly detection algorithms like isolation forest and DBSCAN clusters. But are these the right tools for the job? Or can they leverage Large Language Models (LLMs) to get the task done?
In this post, we’ll explore the possibilities of using anomaly detection algorithms and LLMs to detect bot-generated survey responses. We’ll dive into the strengths and weaknesses of each approach and discuss the potential outcomes.
So, if you’re curious about how to tackle this challenge, keep reading!