The article is very interesting and as another commenter said, easy to read. The impressive number of articles analyzed and the methodology explained allows the reader to have a synthesis of the water research in Latin America and the Caribbean (LAC)

Regarding figures 5 and 9, it is clear that the larger number of publications are produce in Brazil and Mexico. However, those countries are also the LAC countries which more inhabitants, therefore, it would be interesting to see the same result normalized by the population of each country

Regarding figure 10, as a Hydrologist I am proud of that result, which is totally consistent with the definition of Hydrology (Rosbjerg and Rodda, 2019), that supports the relationship of hydrology whit many other disciplines.

An important finding is the description of the research topics presented in figures 1, 6, 7 and 8.  In that sense, it is interesting to verify that statistic methods and water sampling are the predominant methodologies art the articles. Also, I raise the question if that groundwater could be another blind spot at the pacific side of LAC?

Figure 4 produce a feeling of identification for a LAC water science research. It demonstrated that local problems are common to LAC community

Just for discussion and representing those who are not familiar with machine learning I wounder how much difference exists between the results obtaining with the survey and the results obtained with the machine learning methodology. It would be nice to have that chance with the complementary material

Rosbjerg, D. and Rodda, J. 2019 IAHS: a brief history of hydrology. History of Geo- and Space Sciences. Vol 10(1) pp 109-108. https://hgss.copernicus.org/articles/10/109/2019/},

The paper presents a relevant, consistent study on water resources research in Latin America and the Caribbean (LAC). According to the Authors, more than 20000 papers written in Portuguese, Spanish and English were analyzed. I would like to congratulate the Authors for the enormous effort, enabled using a powerful tool, “Machine Learning”. The presented methodology is well structured and is useful for other areas of knowledge. The results and their interpretation support the presence of bright and blind research spots in LAC, indicating where and what could be developed and improved in terms of research and networks of the water community. One notes that collaboration among Brazil and other LAC countries should be augmented. Moreover, it is of concern the lack of collaboration to study international basins and the delicate results that affect all the countries involved. There are few initiatives like those within LAD-IAHR (Latin America Division of the International Association of Hydro-Environment Engineering and Research) to promote integration in LAC water community and foster Portuguese and Spanish publications, that is important for this area.

Besides this general analysis, some specific comments are pointed out to rather improve the text:

In the Abstract, the first sentence could be improved. Please, state clearly the meaning of “are on particular display”;

In the Introduction section, please cite Fortunato et al. (2018) in the sequence of “Science of Science” in its first appearance. Maybe a brief explanation would also help;

Paragraph (lines 66-70) seems to lack of a main idea…actually, it seems unnecessary to me;

In the Materials section, the use of “our” methodology seems out of place. The steps for corpus collection do not seem an innovation. I would rather write: “The process of corpus collection consisted of four steps”;

In lines 76-79, the Authors are explaining the first step of the method: (i) querying online databases. I suggest not to use “i, ii, iii” again to avoid ambiguity. Use a,b,c or 1,2,3 instead.

In line 87, first sentence, please refer to Equations (1) and (2).

I suggest performing a clear correspondence of the four steps stated in the first sentence of this section with the rest of it (cohesion and coherence in paragraph writing). If I´m not mistaken:

 Item “(i) querying online database" is explained in one paragraph;

Items "(ii) retrieving documents and (iii) iteratively assessing quality of the corpus and correcting bias" are illustrated in another paragraph;

The next paragraph is related to item (iii);

Item "(iv) cleaning the corpus" is explained in the last paragraph.

In subitem 2.2, why these databases were chosen?

In Methods section, please correct: we detailed.

In line 167: The general metadata corresponds;

In the Results section, the first paragraph seems to lack of a main idea…I would rewrite it. Please make the correspondence to the respective Figure you are describing;

In lines 285 and 289, please explain the abbreviations in their first appearance in the text;

In lines 314 and 315, please rewrite this first sentence;

In line 349, please break the sentence into several sentences to improve its understanding;

In lines 375-380, I was confused with the text. Could you rewrite it, please?

In subsection 5.5, due to the existence of a lot of information, I was also confused with the analysis and correspondence (in particular) to Figure 10b. Could the Authors further clarify this correspondence in the Figure?