We recommend focusing your time on improving the low performing tags, as this will have a bigger impact on increasing your classifier performance.
For identifying low performing tags on your classifier, go to Build > Stats, and explore the Precision and Recall values for each of your tags:
In the example above, the tag Advertisement has high precision and recall: 98%.
In contrast, the tag Security & Privacy has slightly worse precision and recall:
In this case, you should focus your work on improving the Security & Privacy tag before focusing on boosting the Advertisement tag.
Moreover, assuming you value precision and recall equally, in this case, you first should focus on improving the recall of Security & Privacy to reduce the number of false negatives in this tag. After improving this metric, you could focus on boosting the precision of this tag to bring down the number of false positives.