A Phrase-Level User Requests Mining Approach in Mobile Application Reviews: Concept, Framework, and Operation
Round 1
Reviewer 1 Report
Original and interesting research, good job.
Author Response
Thanks for your comments.
Reviewer 2 Report
There are numerous editorial errors in the inconsistencies of the formats of the References on pages 21 and 22 and 23. These inconsistencies are in capitalization of and use of italics in titles. For example, please note the following:
1.) References 7, 9, 15, 18, 30, 37, 38, 41, 45, 47, 51 and 54 use no italics for titles of conference Proceedings and Journals while the following References do use italics: 10, 11, 12, 13, 14, 19, 22, 24, 25, 29, 31, 35, 39, 42, , 52, 53, etc. I believe that all titles of Journals and Conference Proceedings need to be in italics with capitalization of all words except for prepositions such as "for" and "the", "on", "of", etc. Reference 15 does not use capitalization for any words in title except for IEEE.
2.) References 40, 45, 50 and 52 and 54 do not include page numbers of article cited and there are many more References that also are incomplete without specific pagination of reference cited.
3.) Reference 54 also does not provide publisher of work cited as "Lecture Notes in Computer Science"
This submission needs a very careful editorial check of each and every one of the References cited on pages 21 and 22 and 23.
Author Response
Point 1: There are numerous editorial errors in the inconsistencies of the formats of the References on pages 21 and 22 and 23. These inconsistencies are in capitalization of and use of italics in titles.
Response: We have carefully checked the external BibTeX file, especially for the capitalizations. As there are changes and adds for the bibliographies, the number may not match the previous version. Thanks for your detailed statement.
Point 1: References 7, 9, 15, 18, 30, 37, 38, 41, 45, 47, 51 and 54 use no italics for titles of conference Proceedings and Journals while the following References do use italics: 10, 11, 12, 13, 14, 19, 22, 24, 25, 29, 31, 35, 39, 42, , 52, 53, etc. I believe that all titles of Journals and Conference Proceedings need to be in italics with capitalization of all words except for prepositions such as "for" and "the", "on", "of", etc. Reference 15 does not use capitalization for any words in title except for IEEE.
Response: We have checked the published papers for the journal, and it seems that the titles of Journals are in italics while Conference Proceedings are not. As we use the latex template, the italics style is not set manually. Meanwhile, all capitalizations have been fixed especially for the Conferences and Proceedings as 7, 9, 15, 18, 39, 42, 43, etc.
Point 2: References 40, 45, 50 and 52 and 54 do not include page numbers of article cited and there are many more References that also are incomplete without specific pagination of reference cited.
Response: Page numbers have been added in the BibTeX file as 41, 46, 52, 54, 56.
Point 3: Reference 54 also does not provide publisher of work cited as "Lecture Notes in Computer Science".
Response: Reference 54 has been fixed as 56 "Bernard, S.; Heutte, L.; Adam, S. Influence of hyperparameters on random forest accuracy. International Workshop on Multiple Classifier Systems. Springer, 2009, pp. 171–180".
Reviewer 3 Report
App reviews are user comments on their impressions, requirements, and questions posed since using the app. Bug reports and usage requirements evaluated and derived from app feedback fuel app iterations, which is also an issue that app designers and engineers are working to solve. However, because of the large number of software reviews of varying nature, there is still a lot of noise. Manually analyzing software feedback is a complex and time-consuming task. To address this, a novel methodology based on fourteen extracted features is proposed as an automatic tool for predicting high priority user requests. With the support of app changelogs, a semi-automated technique is used to annotate specifications of high or low priority. The viability of the solution is evaluated and the concepts are interpreted using reviews from six applications found on the Apple App Store. The methodology outperforms the IDEA process in terms of precision and recall, with an overall precision of 75.4 percent and recall of 70.4 percent. The method can be used in software creation to help app developers easily define user needs and introduce app management and evolution.
This article has a good structure and I find it easy to be understood by novice readers. Both the abstract and the conclusion have a correlation to the article's content, mathematical formulas, figures and tables and are connected to one another.
More details should be added about the hardware used. Also, figures regarding the architecture and NLP process should be added.
Spell checking is needed "feature work" -> "future work", etc.
More references should be added related to multi-language sentiment analysis.
Author Response
Point 1: More details should be added about the hardware used. Also, figures regarding the architecture and NLP process should be added.
Response: We have added more details for the proposed approach and hardware used in Section 3 and etc. (Line 204, 255, 393, 657) And a figure for the overall architecture and NLP process has been supplemented at the beginning of Section 3. (Figure 2, Line191)
Point 2: Spell checking is needed "feature work" -> "future work", etc.
Response: We have carefully checked the spelling. (Line 107 "praise" -> "praiseful", Line 155 "use"->"uses", Line 161 "is"->"are", Line 422 "date"->"data", Line 493 "less"->"fewer", Line 581 "apply"->"applies", Line 664 "much"->"many, "Line 785 "feature work" -> "future work")
Point 3: More references should be added related to multi-language sentiment analysis.
Response: We have added a subsection about sentiment analysis in Section 2 Related work. (Line 112)
