Review History
Event classification from the Urdu language text on social media

The current revision addresses the minor revisions suggested in the final round and it can now be accepted for publication in its current form.

I took over handling your manuscript as the previous Academic Editor became unavailable.

While the content of the paper is deemed scientifically correct, the writing needs to improve that the contributions and the findings are clearly stated throughout, particularly in the abstract, introduction and conclusion. Please revise these sections to make this clear.

Please proofread the paper to improve its readability.

Thank you for your latest version.

Unfortunately, you have not adequately addressed the critiques. Please note that when submitting a rebuttal to a critique from editors or reviewers it is not acceptable to say "All the recommendation are incorporated in the article." and provide the track changes version. You should instead offer a point by point response to all concerns raised, highlighting which sections were added or taken out. This is particularly important if the reviewer or editors have raised serious concerns.

It is also problematic that the whole analysis seems to have changed, some of the tables in this new version appear to be different, there is no explanation of the changes.

In this case it appears you have only partially addressed the concerns raised. You have addressed minor issues by subtracting text but you ignored or made minor efforts to address the major issues. Please read my summary from the previous round, and address each concern in detail.

Frankly, some of the changes raise suspicions. Previously a reviewer had concerns that using Title alone had unreasonable accuracy, and I gave you a few reasons why this might be the case. The correct thing to do here was to address these concerns. Instead you seem to have simply dropped this table in the results.

In order to be published, major revisions are required in particular
see Reviewer 2's comments about lack of detail in results.

In multiple sections, you say your dataset consists of more than one million sentences/instances. But the actual number seems to be 0.1 million. Also do not write "1,02,962". The comma is in the wrong place, and it misleads the reader into thinking there are one million, rather than one-tenth that. Write "102,962" instead.

You also do not report the number of distinct posts. A single post has
one title and many sentences. As reviewer 1 points out, you report
unreasonably high accuracy when using Title as a feature. I suspect
that when you split the dataset into training and test sets, you
are splitting by sentence and not by post. It is therefore not
surprising that you can predict with 100% accuracy based on the title!

If this is true you need to redo the analysis to ensure that the
features of the post do not end up polluting your results.

I think your reporting of the results needs improving. Figures 4-6 are
redundant with the tables. You should combine these into a smaller
number of tables or figures. But this point is secondary if the
evaluation is flawed.

I also ask that you improve your literature review section. In particular, you say:

> Urdu textual contents explored [27] for classification using the majority voting algorithm. They categorized Urdu text into seven classes i.e., Health, Business, Entertainment, Science, Culture, Sports, and Wired. They used 21769 news documents for classification and reported 94% precision and recall. Dataset evaluated using these algorithms, Linear SGD, Bernoulli Naïve Bayes, Linear SVM, Naïve Bayes, random forest classifier, and Multinomial Naïve Bayes. Textual classification is close to our problem that is events classification by text at the sentence level, but it is completely different. They did not report the overall accuracy of the system for multiple classes. The information about feature selection is also omitted by the researchers but comparatively, we disclosed the feature selection, engineering, and accuracy of classifiers for multi-classes. Our dataset set consists of 1,02,960 instances of sentences and twelve (12) classes that are comparatively very greater

Citation [27] (Daud et al) is a survey, summarizing the current state
of the art. When you say "they reported 94% precision and recall" and
"They did not report the overall accuracy of the system" this is
misleading. Daud is summarizing the results of Sajjad and Schmid
(2009) (amongst many others). It is not appropriate for you to
criticize them for not reporting all statistics, as it is a review
paper.

I recommend that you take this section out, and instead cite the
primary research in the papers cited in Daud et al, and compare your
work against the primary research works.

[# PeerJ Staff Note: Please ensure that all review comments are addressed in a response letter and any edits or clarifications mentioned in the letter are also inserted into the revised manuscript where appropriate.  It is a common mistake to address reviewer questions in the response letter but not in the revised manuscript. If a reviewer raised a question then your readers will probably have the same question so you should ensure that the manuscript can stand alone without the response letter.  Directions on how to prepare a response letter can be found at: https://peerj.com/benefits/academic-rebuttal-letters/ #]

In order to be published you should address all issues identified by the reviewers. In particular it is very important to be clear about the nature of the work in the title and abstract, and to describe all methods, all algorithms used, and to be clear about the size of the training set.

[# PeerJ Staff Note: Please ensure that all review comments are addressed in a rebuttal letter and any edits or clarifications mentioned in the letter are also inserted into the revised manuscript where appropriate.  It is a common mistake to address reviewer questions in the rebuttal letter but not in the revised manuscript. If a reviewer raised a question then your readers will probably have the same question so you should ensure that the manuscript can stand alone without the rebuttal letter.  Directions on how to prepare a rebuttal letter can be found at: https://peerj.com/benefits/academic-rebuttal-letters/ #]