Interesting. Data scarcity is a common problem in most data science projects. Am looking forward to the next article how we can generate synthetic conversations which can be useful for training!