What are the steps to implement Machine Learning on rsyslog?

I'm a beginner in Machine Learning and I'm not sure which middleware is suitable for running models on Syslog data. Could you recommend some options for me to consider?