MN-DS: A Multilabeled News Dataset for News Articles Hierarchical Classification
Round 1
Reviewer 1 Report
The manuscript presents a dataset with 10,917 manually labeled news oriented, mainly, to the training of news classification models by topic.
The manuscript is clear and concise. The usefulness of the dataset has been demonstrated by training different text classification models. Although the number of items in the dataset is not very high, it can be useful for the intended purpose.
Author Response
Thank you for your comments.
Reviewer 2 Report
In line 89, a reference seems to be missing ("[?]")
Caption of Figure 2 should state what the black lines are indicating.
Table 2 and Table 3 are too wide and go beyond the page margin.
Is every news article assigned to exactly one category on a certain level, or are some news articles assigned to multiple categories of the same level? This should be stated in the manuscript.
Author Response
Thank you for your comments.
Below you will find each of your comments (numbered), followed by our reply. In each of our replies, we will present reasoning, as well as any actions taken in terms of adjusting the manuscript.
1. In line 89, a reference seems to be missing ("[?]")
Thank you for your comment. We added missing reference to the article.
2. Caption of Figure 2 should state what the black lines are indicating.
Following the reviewer’s suggestion, we have added an explanation that black lines on the chart indicate the 95% confidence intervals.
3. Table 2 and Table 3 are too wide and go beyond the page margin.
Thank you for your comment. We reduced the fonts in these tables for review purposes. The final tables will be properly formatted later in the editorial process.
4. Is every news article assigned to exactly one category on a certain level, or are some news articles assigned to multiple categories of the same level? This should be stated in the manuscript.
Following the reviewer’s suggestion, we clarified this issue in lines 100-106 of the revised manuscript.
Thanks again for your review. We believe that your questions and comments have helped with improving the manuscript.