I'm currently working on a text multi-class classification problem where i have to find the root Cause analysis of the Bug occured in the application.For that i have summary of the error which is text and actual rca.I have around 6-7k entries where each summary is one liner.I have tried Tfidf for representation and every other ML algorithm for classification but accuracy stays around 50-60%.can anyone help me with the Representation and classification technique to use? Also i needed help with Language modelling and how it does help for text classification.P.S-I have 12 classes.
Thanks in Advance!
Have you taken the "NLP in tensorflow" course (third course of the Tensorflow in Practice specialization)?
If not, I recommend you to go through the video lectures, it's helpful.
Try also to know what's the reason to get a low accuracy and check if the model is overfitted on the training set.