A social media-based cross-topic sentiment research method

Abstract

The application provides a cross-topic sentiment research method based on social media, takes the comment emotions of two topics as dependent variables and independent variables respectively, selects a model which can take variable lag period into account, can explore dynamic mechanism, and quotes real comments to analyze sentiment amplification phenomenon; the data screening rules are more comprehensive, data cleaning is more thorough, language range is delimited, short comment influence is excluded, and robot comment interference is excluded; meanwhile, the auxiliary control variable setting of the application is more perfect, and can more comprehensively exclude exogenous factor interference; therefore, it can be seen that the application adopts a dynamic regression model, adds independent variable lag period, further studies the dynamic mechanism between climate change and the sentiment of a disease topic, studies the linkage mechanism of topic sentiment, and provides a favorable clue for cross-topic sentiment research on social media.

Description

Technical Field

[0001] This invention belongs to the field of semantic analysis technology, and in particular relates to a method for cross-issue sentiment research based on social media. Background Technology

[0002] Social media, as a platform for public expression, provides a reference for research on internet platforms. Climate change, a widely discussed topic on social media platforms, allows for real-time monitoring of changes in public attitudes towards this issue by analyzing sentiment shifts.

[0003] Currently, research on the emotional impact of external events on climate change topics mainly includes: (1) the impact of the number of infections and deaths of a disease on the number of posts and comments on climate change during a certain period of disease; and (2) the impact of the number of infections and deaths of a disease on different emotions related to climate change topics during a certain period of disease.

[0004] The above research methods respectively confirm the limited attention pool and limited worry pool hypotheses, namely, during a period of disease outbreak, when the number of infections and deaths from that disease increases, the number of discussions on climate change topics decreases, and the negative sentiment in comments on climate change topics also decreases. However, the above methods only use the simplest least squares regression approach, with the number of infections and deaths from that disease as independent variables, the number and sentiment of climate change comments as dependent variables, and the number of climate change news items and extreme weather events as control variables. This only studies the impact of current exogenous events on current climate change topics.

[0005] Existing research on the factors influencing sentiment on climate change topics on social media during a certain disease primarily focuses on the number of comments on climate change topics and the sentiment surrounding them. The main influencing factors are the number of infections and deaths from the disease. However, this research does not consider the impact of sentiment on climate change topics on the sentiment surrounding the disease, nor does it consider adding lag terms to the regression model to account for historical information. Summary of the Invention

[0006] To address the aforementioned problems, this invention provides a method for cross-issue sentiment research based on social media.

[0007] A cross-issue sentiment research method based on social media includes the following steps: From all the collected comments, comments related to climate change and comments related to a specific disease were selected. After cleaning the selected comments, they were divided into three groups: a climate change group containing only climate change keywords, a disease group containing only disease keywords, and a group that mentions both climate change keywords and disease keywords. The daily scores of the climate change group, a certain disease group, and the co-mentioned group were obtained under eight emotions. The eight emotions are negative emotion 1, negative emotion 2, negative emotion 3, negative emotion 4, positive emotion 1, positive emotion 2, positive emotion 3, and positive emotion 4. Obtain auxiliary control variables, including the number of newly confirmed cases of a certain disease in country A on a single day, the number of newly confirmed cases and deaths of a certain disease in country A on a single day, the disease response tracking index in country A, natural disasters in country A that cause losses exceeding US$1 billion, the dates of speeches by important figures in country A, and the monthly number of reports in country A about climate change or global warming. Based on the daily scores of different groups and auxiliary control variables, three analytical methods and three models were developed: the three methods were: a time series analysis method of the number of climate change comments and the number of deaths from a certain disease; an analysis of variance method for discussing a single topic and discussing two topics simultaneously; and a sentiment association analysis method for comment examples that discuss both climate change and a certain disease. The three models were: a least squares regression model of the number of climate change comments and the number of deaths and infections from a certain disease; a least squares regression model of the sentiment of climate change comments and the number of deaths from a certain disease; and an autoregressive distributed lag error correction model for discussing only climate change and only a certain disease. Based on three analytical methods and three analytical models, this study analyzes the factors influencing the number of comments on climate change topics on social media during a certain disease period, the factors influencing the sentiment of climate change topic comments, the relationship between the sentiment of comments on climate change, a certain disease, and comments discussing both simultaneously, the sentiment correlation between climate change topics and a certain disease topic, and presents example topics.

[0008] Furthermore, regular expressions are used to filter out comments related to climate change and comments related to a specific disease; Furthermore, the method for cleaning the selected comments is as follows: 1) Keep English comments; 2) Delete duplicate comments, including those that are identical in content but differ only in capitalization; 3) Delete comments containing information related to bots; 4) Delete comments containing usernames or information related to bots; 5) Only retain comments with more than 15 words.

[0009] Furthermore, the method for obtaining the daily score of any set of comments under any sentiment is as follows: Obtain the total number of words in each comment in the current group and the number of words containing the current sentiment in each comment. Use the ratio of the number of sentiment words in each comment to the total number of words as the sentiment intensity of each comment under the current sentiment. The average emotional intensity of all comments in the current group on that day under the current emotional state is used as the daily score of the current group of comments under the current emotional state.

[0010] Furthermore, the method for constructing a least-squares regression model of climate change commentary with the number of deaths and infections of a certain disease is as follows: Using the number of climate change comments as the dependent variable, the number of deaths or infections of a certain disease as the independent variable, and the disease response tracking index of country A, the monthly number of reports on climate change or global warming in country A, natural disasters in country A that cause losses of more than US$1 billion, and the dates of speeches by important figures in country A as control variables, the least squares regression coefficients of climate change comments and the number of deaths and infections of a certain disease were obtained.

[0011] Furthermore, the method for constructing a least-squares regression model between climate change commentary sentiment and the number of deaths from a certain disease is as follows: Using eight different emotions as dependent variables and the number of deaths from a certain disease as independent variables, the disease response tracking index of country A, the monthly number of reports on climate change or global warming in country A, natural disasters in country A that cause losses exceeding US$1 billion, and the dates of speeches by important figures in country A were used as control variables to obtain the least squares regression coefficient between climate change commentary sentiment and the number of deaths from a certain disease.

[0012] Furthermore, the following models are constructed from two perspectives: one focusing solely on climate change, and the other focusing solely on a specific disease, to correct the autoregressive lag error of the distribution. Direction 1: Climate change comment sentiment as the dependent variable, disease comment sentiment as the independent variable, and the disease response tracking index of country A, the monthly number of reports on climate change or global warming in country A, natural disasters in country A that cause losses of more than US$1 billion, and the dates of speeches by important figures in country A as control variables. Direction 2: Using sentiment in comments about a certain disease as the dependent variable and sentiment in comments about climate change as the independent variable, the disease response tracking index of country A, the monthly number of reports about climate change or global warming in country A, and the dates of speeches by important figures in country A during natural disasters that cause losses exceeding US$1 billion in country A are used as control variables.

[0013] Furthermore, based on a least-squares regression model analyzing the number of climate change comments on social media during a certain disease period, and relating them to the number of deaths and infections of a particular disease, the following factors were identified: An increase in the number of deaths from a certain disease significantly reduced the number of discussions on climate change, while an increase in the number of infections from a certain disease had no significant impact on the number of comments on climate change. Based on a least-squares regression model of climate change sentiment and the number of deaths from a certain disease, the following factors influenced sentiment on social media regarding climate change topics during a certain disease period: An increase in the number of deaths from a certain disease will lead to a decrease in negative sentiment 2, negative sentiment 1, negative sentiment 3, and negative sentiment 4 in comments on climate change topics, and an increase in positive sentiment 1 and positive sentiment 3 in comments on climate change topics, while having no significant effect on positive sentiment 2 and positive sentiment 4 in comments on climate change topics. The following analysis of the sentiment relationship between comments on climate change, the disease itself, and simultaneous discussions of both topics on social media during a specific disease period was conducted using ANOVA: Comments discussing both climate change and a disease showed significantly higher levels of negative sentiment (1, 2, 3, and 4) compared to comments discussing only climate change or only a disease, while positive sentiment (1, 2, 3, and 4) did not exhibit these characteristics. Based on the analysis of autoregressive distributed lag error correction models that only discuss climate change and only discuss a specific disease, the sentiment correlation between climate change topics and disease topics on social media during the period of a specific disease is as follows: The current changes in negative sentiment 1, negative sentiment 2, negative sentiment 3, and negative sentiment 4 and positive sentiment 3 in comments on a certain disease topic will positively influence the corresponding sentiment in comments on a certain climate change topic, and vice versa. An increase in negative sentiment 1, negative sentiment 2, and negative sentiment 4 in comments on a certain disease topic will lead to a long-term increase in the corresponding sentiment in comments on climate change topics, and vice versa. An increase in negative sentiment 1, negative sentiment 2, and negative sentiment 4 in comments on climate change topics will lead to a long-term increase in the corresponding sentiment in comments on a certain disease topic. Furthermore, based on example comments discussing both climate change and a certain disease, the sentiment association analysis method presents three example topics as follows: Theme 1: Comments that simultaneously convey the crisis and its severity, and the emotions contained in the comments include negative emotion 2, negative emotion 3, and negative emotion 4; Theme 2: Commentary showcasing the crisis of collective action and scientific authority, with the commentary containing both negative sentiment 1 and positive sentiment 3; Theme 3: Showcase comments expressing hope and expectations for the future, with the comments including positive emotions 1, 2, and 4.

[0014] Beneficial effects: 1. This invention provides a method for cross-topic sentiment research on social media. It uses the sentiment of comments on two topics as dependent and independent variables, respectively, and selects a model that incorporates the lag period of the variables to explore dynamic mechanisms. It also uses real comments to analyze the amplification of sentiment. Furthermore, the auxiliary control variables in this invention are more comprehensively set, enabling the elimination of interference from exogenous factors. Therefore, this invention employs a dynamic regression model, incorporating the lag period of the independent variables, to further study the dynamic mechanism between climate change and sentiment on a specific disease topic, and to investigate the linkage mechanism of topic sentiment, providing valuable clues for cross-topic sentiment research on social media.

[0015] 2. This invention provides a cross-issue sentiment research method based on social media, with more comprehensive data screening rules, more thorough data cleaning, defined language range, elimination of the influence of short comments, and elimination of interference from robot comments. Attached Figure Description

[0016] Figure 1 A flowchart illustrating a cross-issue sentiment research method based on social media provided by this invention; Figure 2 A time-series descriptive analysis diagram of the number of climate change comments and the number of deaths from a certain disease provided by this invention; Figure 3 This invention provides a schematic diagram of the regression coefficients for the number of deaths and infections of a certain disease. Figure 4 This invention provides a schematic diagram of the regression coefficients for the number of deaths from a certain disease. Figure 5 The violin plot is obtained by performing an analysis of variance on the time series of eight emotion climate change groups, a disease group, and a common mention group provided by this invention. Figure 6 Heat maps of short-term and long-term coefficients provided for this invention; Figure 7 This is a schematic diagram illustrating the subject matter of the present invention. Detailed Implementation

[0017] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0018] To clarify the impact of a disease's mortality rate on the number of climate change comments, the impact of a disease's mortality rate on the sentiment of climate change comments, the mutual influence between the sentiment of disease-related comments and the sentiment of climate change comments, the amplification effect of comparing the sentiment of comments discussing two topics together with the sentiment of comments discussing a single topic, and to conduct a thematic analysis of comments discussing two topics from different sentiment perspectives, this invention provides a cross-issue sentiment research method based on social media, such as... Figure 1 As shown, the specific steps include: Step 1: Filter out comments related to climate change and comments related to a specific disease from all comments collected in one day. After cleaning the filtered comments, divide them into three groups: a climate change group containing only climate change keywords, a disease group containing only disease keywords, and a group that mentions both climate change keywords and disease keywords. It should be noted that this invention uses Python code to filter all collected comments, and uses regular expressions to filter comments related to climate change and certain diseases, dividing them into three groups: those containing only climate change keywords are divided into the climate change group, those containing only disease keywords are divided into the disease group, and those containing both topic keywords are divided into the common mention group.

[0019] The filtered data is cleaned, primarily to remove duplicate comments and bot replies, based on the following principles: 1) Keep English comments 2) Delete duplicate comments, including those that are identical in content but differ only in capitalization. Delete comments containing information about robots. 3) Filter comments containing user names with robot-related information. 4) Only retain comments with more than 15 words.

[0020] Step 2: Obtain the daily scores of the climate change group, a certain disease group, and the co-mentioned group under 8 emotions. The eight emotions are negative emotion 1, negative emotion 2, negative emotion 3, negative emotion 4, positive emotion 1, positive emotion 2, positive emotion 3, and positive emotion 4. It should be noted that this invention uses the NRClixicon dictionary to calculate the number of eight sentiment words, and divides the number of corresponding sentiment words by the number of words in the comment to obtain the sentiment intensity of a single comment; for example, if a comment has a total of 30 words, of which 6 words represent negative sentiment 1, then the intensity of negative sentiment 1 in this comment is 6 / 30=0.2; the mean of the sentiment intensity of all comments in a day is calculated, and a total of 24 daily scores are obtained for the eight sentiments of the climate change group, a certain disease group, and the co-mentioned group.

[0021] Step 3: Obtain auxiliary control variables, including the number of newly confirmed cases of a certain disease in country A on a single day, the number of newly confirmed cases and deaths of a certain disease in country A on a single day, the disease response tracking index in country A, natural disasters in country A that cause losses exceeding US$1 billion, the dates of important speeches by people in country A, and the monthly number of reports in country A about climate change or global warming. Step 4: Based on the daily scores of different groups and auxiliary control variables, conduct three analytical methods and three models: The three methods are: time series analysis of the number of climate change comments and the number of deaths from a certain disease; analysis of variance for discussing a single topic and discussing two topics simultaneously; and sentiment analysis of example sentences discussing climate change and a certain disease. The three models are: least squares regression model of the number of climate change comments and the number of deaths and infections from a certain disease; least squares regression model of climate change comment sentiment and the number of deaths from a certain disease; and autoregressive distributed lag error correction model for discussing only climate change and only a certain disease. Step 5: Based on the three analytical methods and three analytical models, analyze the factors influencing the number of comments on the topic of climate change on social media during a certain disease period, the factors influencing the sentiment of the topic of climate change, the relationship between the sentiment of comments on climate change, a certain disease, and simultaneous discussions of the two, the sentiment correlation between comments on the topic of climate change and a certain disease, and provide example sentences.

[0022] Specifically, such as Figure 2 The diagram illustrates a time-series descriptive analysis of the number of climate change comments and the number of deaths from a certain disease, as provided by this invention. This invention uses the number of climate change comments as the dependent variable, the number of deaths or infections from a certain disease as the independent variable, and the disease response tracking index of country A, the monthly number of reports on climate change or global warming in country A, natural disasters in country A causing losses exceeding $1 billion, and the dates of speeches by important figures in country A as control variables. The least squares regression coefficients of climate change comments and the number of deaths and infections from a certain disease are obtained, thus completing the construction of a least squares regression model for climate change comments and the number of deaths and infections from a certain disease. Figure 3 As shown, the regression coefficients for the number of deaths and infections of a certain disease are presented. It can be seen that an increase in the number of deaths from a certain disease significantly reduces the number of discussions on climate change, while an increase in the number of infections of a certain disease has no significant impact on the number of comments on climate change.

[0023] Furthermore, using eight different emotions as dependent variables, the number of deaths from a certain disease as independent variables, and the disease response tracking index of country A, the monthly number of reports on climate change or global warming in country A, natural disasters in country A that cause losses exceeding US$1 billion, and the dates of speeches by important figures in country A as control variables, the least squares regression coefficients of climate change comment sentiment and the number of deaths from a certain disease were obtained, thus completing the construction of the least squares regression model of climate change comment sentiment and the number of deaths from a certain disease. For example, taking negative sentiment 1 as an example, the dependent variable is negative sentiment 1 in climate change comments, the independent variable is the number of deaths from a certain disease on that day, and the control variables are the same as above.

[0024] like Figure 4 As shown, the regression coefficients for the number of deaths from a certain disease lead to the following conclusions: An increase in the number of deaths from a certain disease will lead to a decrease in negative sentiment 1, negative sentiment 2, negative sentiment 3, and negative sentiment 4 in comments on climate change topics, and an increase in positive sentiment 1 and positive sentiment 3 in comments on climate change topics, while having no significant effect on positive sentiment 2 and positive sentiment 4 in comments on climate change topics.

[0025] Furthermore, this invention performs variance analysis on the time series data of eight emotion-related climate change groups, a certain disease group, and a co-mentioned group, such as... Figure 5 The image shown is a violin diagram. The following conclusions can be drawn: negative sentiment 1, negative sentiment 2, negative sentiment 3, and negative sentiment 4 in comments that discuss both climate change and a certain disease are significantly higher than those that discuss only climate change or only a certain disease. Positive sentiment 1, positive sentiment 2, positive sentiment 3, and positive sentiment 4 do not exhibit the above characteristics.

[0026] Furthermore, in order to study the relationship between the emotions expressed when discussing climate change and when discussing a certain disease, this invention constructs autoregressive distributed lag error correction models from two directions: one for discussing only climate change and the other for discussing only a certain disease, as follows: Direction 1: Climate change comment sentiment as the dependent variable, disease comment sentiment as the independent variable, and the disease response tracking index of country A, the monthly number of reports on climate change or global warming in country A, natural disasters in country A that cause losses of more than US$1 billion, and the dates of speeches by important figures in country A as control variables. Direction 2: Using sentiment in comments about a certain disease as the dependent variable and sentiment in comments about climate change as the independent variable, the disease response tracking index of country A, the monthly number of reports about climate change or global warming in country A, and the dates of speeches by important figures in country A during natural disasters that cause losses exceeding US$1 billion in country A are used as control variables.

[0027] like Figure 6 The image shows a heatmap of short-term and long-term coefficients. Figure 6Only coefficients that are significant at at least a 5% confidence level are retained; insignificant coefficients are left blank. Short-term coefficients are the regression coefficients of the current change in the independent variable, while long-term coefficients are the long-term effects of a one-unit change in the independent variable on the dependent variable. The following conclusions are drawn: 1. The current changes in negative sentiment 1, negative sentiment 2, negative sentiment 3, negative sentiment 4 and positive sentiment 3 in comments on a certain disease topic will positively influence the corresponding sentiment in comments on a certain climate change topic, and vice versa. The current changes in negative sentiment 1, negative sentiment 2, negative sentiment 3, negative sentiment 4 and positive sentiment 3 in comments on a certain climate change topic will positively influence the corresponding sentiment in comments on a certain disease topic. 2. An increase in negative sentiment 1, negative sentiment 2, and negative sentiment 4 in comments on a certain disease topic will lead to a long-term increase in the corresponding sentiment in comments on climate change topics, and vice versa. An increase in negative sentiment 1, negative sentiment 2, and negative sentiment 4 in comments on climate change topics will lead to a long-term increase in the corresponding sentiment in comments on a certain disease topic.

[0028] Furthermore, in order to analyze comments that simultaneously discuss climate change and a certain disease, the two topics are linked together, such as... Figure 7 As shown, this invention selects example sentences from comments that simultaneously discuss two topics, categorizes them into eight emotions, and displays the example sentences, summarizing them into three themes as follows: Theme 1: Comments that simultaneously convey the crisis and its severity, and the emotions contained in the comments include negative emotion 2, negative emotion 3, and negative emotion 4; Theme 2: Commentary showcasing the crisis of collective action and scientific authority, with the commentary containing both negative sentiment 1 and positive sentiment 3; Theme 3: Showcase comments expressing hope and expectations for the future, with the comments including positive emotions 1, 2, and 4. In summary, this invention provides a method for cross-topic sentiment research on social media that can clarify the impact of the number of deaths from a certain disease on the number of climate change comments, clarify the impact of the number of deaths from a certain disease on the sentiment of climate change comments, clarify the mutual influence between the sentiment of comments on a certain disease and the sentiment of climate change comments, clarify the amplification effect of the sentiment of comments discussing two topics compared with the sentiment of comments on a single topic, and complete thematic analysis of comments discussing two topics from different sentiment perspectives, providing valuable clues for cross-topic sentiment research on social media.

[0029] Of course, the present invention may have other various embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art can make various corresponding changes and modifications according to the present invention, but these corresponding changes and modifications should all fall within the protection scope of the appended claims.

Claims

1. A method for cross-issue sentiment research based on social media, characterized in that, Includes the following steps: From all the collected comments, comments related to climate change and comments related to a specific disease were selected. After cleaning the selected comments, they were divided into three groups: a climate change group containing only climate change keywords, a disease group containing only disease keywords, and a group that mentions both climate change keywords and disease keywords. The daily scores of the climate change group, a certain disease group, and the co-mentioned group were obtained under eight emotions. The eight emotions are negative emotion 1, negative emotion 2, negative emotion 3, negative emotion 4, positive emotion 1, positive emotion 2, positive emotion 3, and positive emotion 4. Obtain auxiliary control variables, including the number of newly confirmed cases of a certain disease in country A on a single day, the number of newly confirmed cases and deaths of a certain disease in country A on a single day, the disease response tracking index in country A, natural disasters in country A that cause losses exceeding US$1 billion, the dates of speeches by important figures in country A, and the monthly number of reports in country A about climate change or global warming. Based on the daily scores of different groups and auxiliary control variables, three analytical methods and three models were developed: the three methods were: a time series analysis method of the number of climate change comments and the number of deaths from a certain disease; an analysis of variance method for discussing a single topic and discussing two topics simultaneously; and a sentiment association analysis method for comment examples that discuss both climate change and a certain disease. The three models were: a least squares regression model of the number of climate change comments and the number of deaths and infections from a certain disease; a least squares regression model of the sentiment of climate change comments and the number of deaths from a certain disease; and an autoregressive distributed lag error correction model for discussing only climate change and only a certain disease. Based on three analytical methods and three analytical models, this study analyzes the factors influencing the number of comments on climate change topics on social media during a certain disease period, the factors influencing the sentiment of climate change topic comments, the relationship between the sentiment of comments on climate change, a certain disease, and comments discussing both simultaneously, the sentiment correlation between climate change topics and a certain disease topic, and presents example topics.

2. The method for cross-issue sentiment research based on social media as described in claim 1, characterized in that, Use regular expressions to filter comments related to climate change and comments related to a specific disease.

3. The method for cross-issue sentiment research based on social media as described in claim 1, characterized in that, The method for cleaning the filtered comments is as follows: 1) Keep English comments; 2) Delete duplicate comments, including those that are identical in content but differ only in capitalization; 3) Delete comments containing information related to bots; 4) Delete comments containing usernames or information related to bots; 5) Only retain comments with more than 15 words.

4. The method for cross-issue sentiment research based on social media as described in claim 1, characterized in that, The method for obtaining the daily score of any set of comments under any sentiment is as follows: Obtain the total number of words in each comment in the current group and the number of words containing the current sentiment in each comment. Use the ratio of the number of sentiment words in each comment to the total number of words as the sentiment intensity of each comment under the current sentiment. The average emotional intensity of all comments in the current group on that day under the current emotional state is used as the daily score of the current group of comments under the current emotional state.

5. The method for cross-issue sentiment research based on social media as described in claim 1, characterized in that, The method for constructing a least-squares regression model of climate change commentary with the number of deaths and infections of a certain disease is as follows: Using the number of climate change comments as the dependent variable, the number of deaths or infections of a certain disease as the independent variable, and the disease response tracking index of country A, the monthly number of reports on climate change or global warming in country A, natural disasters in country A that cause losses of more than US$1 billion, and the dates of speeches by important figures in country A as control variables, the least squares regression coefficients of climate change comments and the number of deaths and infections of a certain disease were obtained.

6. The method for cross-issue sentiment research based on social media as described in claim 1, characterized in that, The method for constructing a least-squares regression model of climate change sentiment and the number of deaths from a certain disease is as follows: Using eight different emotions as dependent variables and the number of deaths from a certain disease as independent variables, the disease response tracking index of country A, the monthly number of reports on climate change or global warming in country A, natural disasters in country A that cause losses exceeding US$1 billion, and the dates of speeches by important figures in country A were used as control variables to obtain the least squares regression coefficient between climate change commentary sentiment and the number of deaths from a certain disease.

7. The method for cross-issue sentiment research based on social media as described in claim 1, characterized in that, The following are two separate autoregressive distributed lag error correction models, one considering only climate change and the other considering only a specific disease: Direction 1: Climate change comment sentiment as the dependent variable, disease comment sentiment as the independent variable, and the disease response tracking index of country A, the monthly number of reports on climate change or global warming in country A, natural disasters in country A that cause losses of more than US$1 billion, and the dates of speeches by important figures in country A as control variables. Direction 2: Using sentiment in comments about a certain disease as the dependent variable and sentiment in comments about climate change as the independent variable, the disease response tracking index of country A, the monthly number of reports about climate change or global warming in country A, the dates of natural disasters in country A that caused losses exceeding US$1 billion, and the dates of speeches by important figures in country A are used as control variables.

8. The method for cross-issue sentiment research based on social media as described in claim 1, characterized in that, The following analysis, based on a least-squares regression model of the number of climate change comments and the number of deaths and infections related to a certain disease, reveals the following factors influencing the number of climate change-related comments on social media during a specific disease period: An increase in the number of deaths from a certain disease significantly reduced the number of discussions on climate change, while an increase in the number of infections from a certain disease had no significant impact on the number of comments on climate change. Based on a least-squares regression model of climate change sentiment and the number of deaths from a certain disease, the following factors influenced sentiment on social media regarding climate change topics during a certain disease period: An increase in the number of deaths from a certain disease will lead to a decrease in negative sentiment 2, negative sentiment 1, negative sentiment 3, and negative sentiment 4 in comments on climate change topics, and an increase in positive sentiment 1 and positive sentiment 3 in comments on climate change topics, but will have no significant effect on positive sentiment 4 and positive sentiment 2 in comments on climate change topics. The following analysis of the sentiment relationship between comments on climate change, the disease itself, and simultaneous discussions of both topics on social media during a specific disease period was conducted using ANOVA: Comments discussing both climate change and a disease showed significantly higher levels of negative sentiment (1), negative sentiment (2), negative sentiment (3), and negative sentiment (4) compared to comments discussing only climate change or only a disease. Positive sentiment (1), positive sentiment (2), positive sentiment (3), and positive sentiment (4) did not exhibit these characteristics. Based on the analysis of autoregressive distributed lag error correction models that only discuss climate change and only discuss a specific disease, the sentiment correlation between climate change topics and disease topics on social media during the period of a specific disease is as follows: The current changes in negative sentiment 1, negative sentiment 2, negative sentiment 3, and negative sentiment 4, and positive sentiment 3 in comments on a certain disease topic will positively influence the corresponding sentiment in comments on a certain climate change topic, and vice versa. An increase in negative sentiment 1, negative sentiment 2, and negative sentiment 4 in comments on a certain disease topic will lead to a long-term increase in the corresponding sentiment in comments on climate change topics, and vice versa. An increase in negative sentiment 1, negative sentiment 2, and negative sentiment 4 in comments on climate change topics will lead to a long-term increase in the corresponding sentiment in comments on a certain disease topic.

9. The method for cross-issue sentiment research based on social media as described in claim 1, characterized in that, Based on example comments discussing both climate change and a specific disease, three sample topics are presented below using the sentiment association analysis method: Theme 1: Comments that simultaneously convey the crisis and its severity, and the emotions contained in the comments include negative emotion 2, negative emotion 3, and negative emotion 4; Theme 2: Commentary showcasing the crisis of collective action and scientific authority, with the commentary containing both negative sentiment 1 and positive sentiment 3; Theme 3: Showcase comments expressing hope and expectations for the future, with the comments including positive emotions 1, 2, and 4.

Images