Where is the problem? Having big jumps in probabilities is, IMHO, related to a good discrimination of your algorithm : think about the logit function. Well, maybe that I misunderstand your question

My though is that feature extraction is part of data science. Well, it's just an opinion ;-) More seriouly, I'm not talking about easy standard log files, but custom ones from a very specific software. So, that's really extracting information from context « sentences »