Hello - need some help getting the most out of text analytics. We have over 8600 employee comments and want to categorize into high-level topics. Trying to use the "recommendations" as a starting point. As we review some of the recommendations, they aren't relevant or misaligned. As we manually remove from the statement, is there "machine learning" updating the logic for future recommendations?



We have the same scenario for sentiment, we have over 2000 statements that are "no comments" (ie,.Nil, NA,-, ..., etc) that have been assigned either a positive or negative sentiment value. As we manually change the assigned sentiment and score, is there "machine learning" behind the scenes updating the logic and associated assignment to other similar statements? Or do I need to manually change all 2000 statements?



Ideally, once we create our topics, we want to download and apply to the next engagement survey versus starting from scratch. We are hoping that there is some "machine learning" that is applied to improve the sentiment and association with new comments. Is that an unrealistic expectation?



Welcome, all and any recommendations for getting the most out of text analytics.



Thank you!