An Automatic Annotation Method for Discovering Semantic Information of Geographical Locations from Location-Based Social Networks
Round 1
Reviewer 1 Report
Review for the
manuscript 605877
title “An automatic annotation method for discovering semantic information of geographical locations from Location-Based Social Networks“
submitted to ”International Journal of Geoinformation”
The manuscript describes a new method to annotate Points of Interests (PoIs) by combining spatial, temporal and text information. The authors claim the novelty and efficiency of the proposed method compared to the existing ones. Claim of efficiency is based on based on f-score statistical testing. The paper is likely to generate a noticeable interest from readers as it addresses an important topic in current social network analysis.
The study is properly designed and seems to be well-executed. However, the overall quality of the presentation of the material is poor. Acceptance of the paper has to be conditioned on fixing of the following shortcomings:
In many places the manuscript is hard to understand. The paper needs a lot of grammar and style fixes. Pay special attention to the use of words “semantic” and “semantics” all through the text. In most cases you are simply assigning a category to a PoI but “semantics” implies much more than categorization. Semantics is a very broad concept and the use of these terms in the manuscript is often not informative. Provide more details about services like Frickr and others that you have used the data from in your study. This information will be useful for future readers of the paper as on-line service come and go and are forgotten soon. For example, Flickr has changed its business model a few years ago and its newer data will likely not be suitable for such studies. Describe limitations and shortcomings of the proposed algorithm and its potential pitfalls. Is it generalizable to other data? Do you expect the same performance in other regions? How representative is the data that were used for testing? Will it work in languages other than English? Lines 195-196: This is not necessarily true. The date of the photo may have been entered manually and some of the early photos may have a correct date.
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
The paper presents a method for semantic annotation of POIs using a combination of temporal, spatial, and text patterns with an algorithm to associate category with POI.
The research is scientifically solid and the presentation is careful and clear. The paper also improves on results from other previous but relatively current GIScience works.
My only criticism regards the limited notion of semantics in the form of relatively small sets of keywords. I would expect at least a rudimentary relational structuring and larger corpus of words before calling a method as using semantics. Check e.g.:
the notion of semantics in the 'semantic web' literature.
Details:
Line 47: "often": supporting citations missing
Lines 116-128: "semantic" can have different meanings, please clarify
Line 476: status of an institute is not a scientific criterion
Author Response
Please see the attachment.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
This is a review for the revised version of the paper. The authors have addressed most of the issues brought up in the original review and manuscript can be recommended for publication with one minor correction: the word "semantic" was rightfully edited out in most cases but in a few sentences it was replaced with "meaningful information" or "meaningful categories" that does not sound like a good style as "meaningful" is often used as an antonym for "meaningless". In this case "categories" or "labels" would be a better style but I defer the choice of the right term to the author.
Author Response
We agree with the comments of this reviewer that are benefit for our paper. Now the word “meaningful” in the unsuitable expression “meaningful categories” has been deleted in our revised manuscript (please see the line 290). However, as for "meaningful information", we retain the word “meaningful” in the expression “meaningful information” since that the information in check-in data may be meaningful as well as meaningless.
In addition, we make some minor spell check and correct some grammatical errors in the text (please see the line 128,195,250 and 398).
Thanks again for your comments.
