Analysis of Spatial Interaction between Different Food Cultures in South and North China: Practices from People’s Daily Life
Round 1
Reviewer 1 Report
Main question
An interviewed people characterization should be done in order to try responds to some considerations:
e.g. Gender, origin (large or small city, rural or higher urbanizes areas), age, socioeconomic level could play an important role on the people perspective about North-South relationships, just under “food“ perspective. With the actual information the role of the preconception/prejudice into the question answer may be hide. e.g. the difference in the appreciation is in relationship with North-South position, or because in relationship with rural/small city origin respect to the representative food from a large City.
As the authors says (line 58 “… culture is he product of the proactive construction of different social group…” ). Then analysis about gender, origin location… would be important to enhance the results “wealth”.
Other remarks
Line 124-126. 6,128 answers in a community of 26 million daily users of Zhihu platform, really are representative enough?
Line 214. At which level of “sentiments outweigh balance” the authors considered the comment is negative or positive?
Line 257. What is the meaning of the different pattern of hotspot detection: regional level versus province level? This different scale could imply, by itself, different perceptions about North-South considerations?
Line 284. “Some are less relevant to culture, such as climate, plants…, but, again, but if ” line 58 “… culture is the product of the proactive construction of different social group…”, perhaps climate, plants are “redundant topics” that contributed to different cultural perception about North and South. Or these appreciations are from the analysis of the output scores from Table 3 only? If were possible run a cluster analysis of topics data set could show us some “words families” and enhance the interpretation value of the results from topic classification. Perhaps other topic than food could play a relevant significance, or in relationship with food too (reinforcing the food topic).
Section 5.2 Conclusions and future directions
First paragraph is not a conclusion, only a repetition of material and methods issue.
In the same way, second paragraph, really, is not a conclusion, it seems more a reference to support some results more than a conclusion from the research
Author Response
Reviewer#1, Concern # 1: The word "Materials" in the header "IV. Materials and algorithm" is not clear. Please consider different term or remove the word "Materials". e.g. Gender, origin (large or small city, rural or higher urbanizes areas), age, socioeconomic level could play an important role on the people perspective about North-South relationships, just under “food“ perspective. With the actual information the role of the preconception/prejudice into the question answer may be hide. e.g. the difference in the appreciation is in relationship with North-South position, or because in relationship with rural/small city origin respect to the representative food from a large City. As the authors says (line 58 “… culture is he product of the proactive construction of different social group…” ). Then analysis about gender, origin location… would be important to enhance the results “wealth”.
Author response: Thank you for your kind reminding. As described in this paper, we do think that culture is the product of the construction of different social groups. However, the research goal of this paper is not to analyze the cognitive differences in food culture among social groups of different genders, ages, and economic levels and et al., but to focus on people's consensus or comprehensive perspective on the differences between north and south food culture. This means that this paper requires that the data samples obtained must be relatively random, making them relatively broadly representative. The "Zhihu" platform has a very wide range of users, which includes both a sample of high-income people and a sample of low-income people, while ensuring that the sample includes both male and female groups. In addition, it is hard to obtain personal information about users on the platform of “Zhihu”, such as gender and income.
Reviewer#1, Concern # 2: Line 124-126. 6,128 answers in a community of 26 million daily users of Zhihu platform, really are representative enough?
Author response: Thank you for your kind reminding. “Zhihu”data used in this paper was collected in 2017. According to official statistics, the number of “Zhihu” users reached about 100 million in 2017. “Zhihu” is the Q & A community with the largest number of users and the most active daily activities in China. And, so far, this question has been viewed 93,315,823 times, which shows that this is an interesting and popular topic. The respondents are random in gender, income or other aspects, and come from different regions of the country, which guarantees the representativeness of the data to a certain extent. The number of users with such a large base is enough to show that “Zhihu” is widely recognized by people. The descriptions about data characteristics can be find in the section of “study area and data description”.
Reviewer#1, Concern # 3: At which level of “sentiments outweigh balance” the authors considered the comment is negative or positive?
Author response: Thank you for your suggestion. For sentiment feature analysis, this paper uses a mature dictionary library and sentiment analysis model based on natural language processing and analysis technology for sentiment feature analysis in this study. The analysis model selected in this study is “BosonNLP_sentiment_score”. Each record in the dictionary is composed of words and scores. Its words and ratings come from millions of sentiment annotation data from Weibo, news, forums, and other data sources. Combine negative words and degree adverbs to achieve sentiment semantic analysis and comprehensive scoring. When the score is positive, it indicates positive emotions; when the score is negative, it indicates negative emotions; when it is 0, it indicates neutral emotions.
We have added corresponding supplementary explanations in the paper, and hope that will help readers to understand this analysis process more easily.
Reviewer#1, Concern # 4: Line 257. What is the meaning of the different pattern of hotspot detection: regional level versus province level? This different scale could imply, by itself, different perceptions about North-South considerations?
Author response: Indeed, people use different scales when describing regions. For example, some users use the name of a prefecture-level city to indicate location, while others use province names to indicate location, and some users use regional names (such as northeast, northwest, etc.) to indicate location， and such situations are unavoidable in reality. In order to solve this problem, we take prefecture-level cities as the basic analysis unit in the design of rules for natural language analysis. However, when the place names appear as provinces, this article uses provincial capital cities to indicate the provinces where the prefecture-level cities are located. When a place name appears, this paper uses the center (center of gravity) of the area to represent it. When multiple place names with inclusive relationship appear at the same time, choose the lowest level one.
We have added corresponding supplementary explanations in the section 3.2, and hope that will help readers to understand this analysis process more easily.
Reviewer#1, Concern # 5: Line 284. “Some are less relevant to culture, such as climate, plants…, but, again, but if ” line 58 “… culture is the product of the proactive construction of different social group…”, perhaps climate, plants are “redundant topics” that contributed to different cultural perception about North and South. Or these appreciations are from the analysis of the output scores from Table 3 only? If were possible run a cluster analysis of topics data set could show us some “words families” and enhance the interpretation value of the results from topic classification. Perhaps other topic than food could play a relevant significance, or in relationship with food too (reinforcing the food topic).
Author response: Thank you for your kind reminding. The research objective of this paper is what can best represent the differences between North and South cultures. We obtained Table 3 through text semantic cluster analysis. The clustering results show that food-related sub topics and topic words are the most, and much larger than other Number of discussions on topic. Therefore, the topic of dietary data is the main analysis object. In addition, our analytical framework for discussing cultural interactions between the North and the South and their emotional responses is general and can be used to analyze other topics in Table 3. This article is just an example of the most discussed and most concerned dietary cultural differences.
Reviewer#1, Concern # 5: Section 5.2 Conclusions and future directions: First paragraph is not a conclusion, only a repetition of material and methods issue. In the same way, second paragraph, really, is not a conclusion, it seems more a reference to support some results more than a conclusion from the research
Author response: Thank you for your reminder, we have adjusted and modified the conclusion section.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Dear Authors,
The presentation of the discussion can be improve a little bit in order to point out the significance of your important research.
Author Response
Reviewer#2, Concern # 1: The presentation of the discussion can be improve a little bit in order to point out the significance of your important research.
Author response: Thank you for your reminder, We have revised the parts of discussion and conclusion, you can find revised words in the section of 5.1 and 5.2.
Author Response File: Author Response.pdf
Reviewer 3 Report
I'm intrigued to look more into the findings after reading this abstract. The idea of research that examines cultural geography through food is fascinating. It is interesting to look into the context in which natural language processing can detect spatial interaction patterns between two regions (North and South China). I hope that the study will answer the questions posed and provide useful insights into a new perspective for the field of cultural geography.
To improve the quality of the paper I have some comments and suggestions:
The article describes an approach for understanding cultural differences using Natural Language Processing (NLP) method, with a focus on place names, words related to culture, and emotional words. I congratulate the authors for this interesting idea. Nevertheless, the three aspects should be accompanied as a scientific paper by a good storyline.
I understand that the authors would like to "promote" the NLP methods to detect spatial interaction patterns as an effective solution to a new study on cultural geography. However, more scientific support is needed to support the "claim".
A concrete example: after reading the word 'accurately' line 13, I was expecting to read the method of measurement of accuracy, the result of the measurement, etc in a more scientific workflow. I suggest that the authors add a research workflow in section 3 to help readers learn the steps and read the story easily.
Line 13: ...accurately detect the spatial interaction pattern... (how accurate can the pattern be identified by the method? Which criteria do you use for accuracy measurement ? Is your claim backed by references?)
Line 359: Although this paper extracts the place names in the text accurately...(I can not find the arguments behind this sentence. It would be better)
I would suggest the authors add a sub-section about the accuracy or provide more clear and concrete discussion.
Line 55: 这 (remove this text)
Line 173: Table 1... (please tidy up the table, put it in the next page)
I would say that Figure 2 should be shown in a better way because the three maps in figure 2 can't be interpreted. For example, It's hard for me to relate the text in Line 276-277 about a strong hotspot flow pattern of Guangdong with the maps in Figure 2.
The same comments for Figure 4 and Figure 5. In a better size of the representation, the figures can be improved.
It is stated in the abstract that the results deal with cultural conflicts, but there is no concrete or clear argument that addresses conflicts through methods or approaches - problems are resolved fairly for all conflicts, through your strategies or approaches? Perhaps in the discussion section, you could provide an explanation.
In conclusion, the paper is interesting, but in some paragraphs or sentences, it is difficult to follow. My recommendation to authors to find someone outside the authors for reading (and proofreading), and also to figure out which paragraphs can be changed to strengthen reader interpretation.
Author Response
Reviewer#3, Concern # 1: I understand that the authors would like to "promote" the NLP methods to detect spatial interaction patterns as an effective solution to a new study on cultural geography. However, more scientific support is needed to support the "claim".
Author response: Cultural geography focuses on cultural interactions and the cultural conflicts they produce. These conflicts can be positive or negative. To this end, cultural geographers have conducted a series of empirical research based on data such as interviews and questionnaires. The advantage of these data is that corresponding problems can be designed according to the research needs, and the obtained data is structured and can be directly used for analysis and statistics. However, it is difficult to obtain sample data of large spatial range, large data volume, and multiple types at the same time by the above-mentioned data collection methods. Moreover, some interview data may be reluctantly stated by respondents in order to complete specific tasks, and their authenticity and effectiveness cannot be guaranteed. Online question-and-answer platforms like "Zhihu", where users discuss questions voluntarily and take the initiative to express their true feelings. Few people express false feelings on such platforms. The current NLP technology can better extract the required position and semantic information from these discussions. As in this paper, not only can place names that produce cultural interactions be recognized by NLP technology, but also the cultural semantics of interactions. The identified place names can be used as the Origin region and Destination region of the interactive flow, and the extracted semantic information can be used as the subject of the interaction, which constitutes a lot of OD flows, and the cultural interaction or cultural conflict analysis can be carried out by this. This provides a new solution for the study of cultural interaction in cultural geography and the cultural conflicts it produces.
Reviewer#3, Concern # 2: A concrete example: after reading the word 'accurately' line 13, I was expecting to read the method of measurement of accuracy, the result of the measurement, etc in a more scientific workflow. I suggest that the authors add a research workflow in section 3 to help readers learn the steps and read the story easily.
Author response: In the first part of Section 3.2, we detailed the process of Place name relation inference through two examples. Among them, the first example is the idea of extracting place names with inclusion relationships, and the latter example is the idea of extracting place names with parallel relationships. The idea of inferring the place name relationship is designed by us based on the characteristics of spatial semantic information and the research goals of this article, and this design and extraction goal is achieved through current NLP technology. Regarding the accuracy of place name relationship extraction, first of all, NLP technology is very powerful in identifying place names and their related semantic information, which to a certain extent guarantees the accuracy of the cultural interaction pattern extraction in this paper. In addition, we manually identified some sample data and compared it with the results obtained by NLP, and found that the recognition accuracy of NLP is high and meets the basic requirements of this study.
Reviewer#3, Concern # 3: Although this paper extracts the place names in the text accurately...(I can not find the arguments behind this sentence. It would be better)
Author response: Thank you for your reminder, we found that it was not clear enough. We have added supplementary notes to the revised version. Supplementary words are below:
“respondents. Specifically, the processing strategy presented in Section 3.2 is designed to solve this problem. In addition, several nouns that are not place names may also be used to refer to a certain area. For example, Canton refers to Guangdong. These words will also appear as pronouns of place names in the text data. If these words are not in the corpus as place name, then they will be omitted and missed, thereby reducing the accuracy of the analysis results.”
Reviewer#3, Concern # 4: 这 (remove this text)
Author response: Thank you for reminding. We have made corresponding changes in the paper.
Reviewer#3, Concern # 5: I would say that Figure 2 should be shown in a better way because the three maps in figure 2 can't be interpreted. For example, It's hard for me to relate the text in Line 276-277 about a strong hotspot flow pattern of Guangdong with the maps in Figure 2.
Author response: Thank you for your suggestion. We have improved Figure 1 and Figure 2.
Reviewer#3, Concern # 6: The same comments for Figure 4 and Figure 5. In a better size of the representation, the figures can be improved.
Author response: Thank you for your suggestion. We have improved Figure 4 and Figure 5.
Reviewer#3, Concern # 7: It is stated in the abstract that the results deal with cultural conflicts, but there is no concrete or clear argument that addresses conflicts through methods or approaches - problems are resolved fairly for all conflicts, through your strategies or approaches? Perhaps in the discussion section, you could provide an explanation.
Author response: Thanks for your reminding and suggestions. The statement in our summary is a bit lax. We have revised the summary. Our research can find out the factors that cause the cultural conflict, only help to solve the cultural conflict. In addition, we reorganize the conclusion, and further elaborate the relevance of this study and cultural conflict and the contribution of the research results to the theory of cultural conflict.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The changes done improve de manuscript
Reviewer 3 Report
I am pleased with the response and improvements to the paper. Thank you very much for all these changes. The revised paper is now clear and makes a difference.
Therefore, I accept the paper in the current form.
