Hi,

Were the classical Telugu writers aware of the concept of sarcasm? Maybe the British introduced the concept and a vague analogue is now present. When I was employed at the College of Micronesia in Phonpei (formerly Phonapei), Micronesia, none of the students who spoke any the predominant five Micronesian languages  understood or used sarcasm.  If this relates to Telugu neither your rules or a decision tree (even a random forest, a very powerful tool that should be very easy and similar to run instead) would find very consistent results. (However, I did hear that complements were used to lead people on to their disadvantage. A Canadian friend called it "twitting", but it was rarely good-natured. I suppose it was a sort of sarcasm, but with a very restricted application.)

Decision tree results in overfitting the data