Printer Friendly

Research on the crisis of public opinion in the network of social contradictions in the Mongolian-Chinese bilingual large data platform.

1. Introduction

With the advent of the Internet era, the Internet has become a hub for the exchange of ideas, cultural communication and information distribution. Network public opinion gradually become a barometer of social public opinion and weather vanes. However, the tendency of clustering is inevitable in the process of network public opinion dissemination. If once it is malicious manipulation, it will cause great disturbance and harm to social stability and people's life. In such circumstances, the early warning of the network public opinion crisis about social contradictions is particularly important. Accurate and effective evaluation of the crisis level become the prerequisite and key to identify and deal with network public opinion crisis. With the development of the minority culture, the application of the Mongolian people to the Internet is also increasing. Therefore, this study is based on the bilingual identification of Mongolian and Chinese, and is devoted to the evaluation of the level of network public opinion crisis of social contradictions, to explore the influencing factors and evaluation methods of accurate evaluation of crisis level. We strive to find potential public opinion crisis, timely alarm, and make effective response measures.

2. Theory

2.1. Mongolian-Chinese bilingual identification

As a minor ethnic group in China, Mongolians formed their own unique writing system-Mongolian. Mongolian and English, Russian and other phonetic characters are phonemes text. However, it is different from English writing. Moreover, there are many differences between Mongolian and Chinese. For example, Mongolian is phoneme and Chinese is morpheme. The number of Chinese characters is much larger than Mongolian. Mongolian is a word for the unit, there is no kerning; and Chinese characters is the word for the unit, there is no word pitch, only kerning (Hou, 2012).

In order to promote the development of science, education and culture in minority areas, and to broaden the knowledge of minority students, to facilitate their access to a wider range of information, many scholars have done much research on bilingual recognition of Mongolian and Chinese. Wang, Mecklinger, Hofrnann & Weng found that language does not function in semantic accessibility (Wang et al, 2010). HUU You & BAI Xue-jun studied the role of semantic activation in Chinese, and found that both high-frequency and low-frequency vocabularies are earlier than semantic activation and conducive to semantic integration when the stimulus is very short (Hu and Bai, 2012). In the field of Mongolian-Chinese machine translation, CHEN Lei proposed to combine the manually-written sequence rules with the rules of automatic acquisition of Mongolian-Chinese bilingual parallel corpus to reduce the influence of the machine translation of Chinese and Mongolian caused by the size of less than parallel corpus and a large number of long distance transfer (Chen, 2013). SU Chuan-jie used the Mongolian-Chinese bilingual parallel corpus automatic study synchronous context-free grammar to realize Mongolian Chinese machine translation (Su, 2014).

At present, the most widely used software is a series of software developed by Inner Mongolia Mengke Li Software Company. "Mengke Li" specializes in ethnic minority text information, digital, network technology development and services. Therefore, the Mongolian people can use this kind of software to access the Internet information, express their attitudes and perspectives and enrich the cultural life.

2.2. Social contradictions and early warning

The social contradiction is the situation of mutual promotion, mutual complement and mutual development, and there is opposition, mutual exclusion andmutual struggle between different social groups or social classes in a social community (Wu, 2015). The early warning mechanism of social contradiction is to find out the inherent law by analyzing the existent or potential social contradictions, and then select the corresponding evaluation index, establish the social contradiction early warning system. Finally, through the accurate identification and tracking of social contradictions, we can judge the network public opinion crisis of social contradictions as early as possible so as to minimize the crisis level and adverse effects. Through the analysis of network public opinion on the information resources of cyberspace communication, we can actively divert the unexpected situation and issue early warning information to the relevant administrative departments, so as to provide sufficient information support for decision-making.

3. Literature review

Internet anonymity and interaction makes the rapid development of network public opinion, the network public opinion research has also been widespread concern of scholars at home and abroad. In the construction of the network public opinion early warning index, Zeng Run-xi & XU Xiao-lin thought that the detection, the collection, the analysis, the warning, the precontrol, five subsystems constitute the network public opinion unexpected event early warning mechanism (Zeng and Xu, 2009). They use the Analytic Hierarchy Process (AHP) to construct the warning index system of network public opinion emergencies, which are warning source, warning sign and police intelligence, and weight each index according to its importance (Zeng and Xu, 2010).

ZHANG Yi-wen & QI Jia-yin aiming at the characteristics of unconventional events, this paper constructs an index system to measure and evaluate the public opinion hotspots of unconventional emergencies, and to clarify the deep-seated causes of public opinion fluctuation (Zhang and Qi, 2010). DAI Yuan & HAO Xiao-wei constructs the network public opinion security evaluation index system from the four dimensions of spread, public attention, content sensitivity and attitude tendency (Dai and Hao, 2010). Chen Yue & Li Ling-chao introduced the interval number to customize the quantitative indicators, using AHP method to determine the weights of indicators at all levels, and the establishment of the network public opinion threat assessment and ranking model based on the TOPSIS method (Chen and Li, 2012). Shi Gui-yun & Zhao Hai-yan conducted a risk analysis on the urban-rural fringe in the areas of economy, environment and government governance, and sorted out the main social risks in the urban-rural fringe, and set up the "social risk early-warning index system" (Shi and Zhao, 2012). Wang Lin, Huang Li-fang using the causal relationship forcing the method found that social contradictions can be attributed to the various subsystems of social instability. They proposed to establish the quantitative analysis model of each subsystem of the society, to determine the weight of each index system, to construct the social contradiction early warning index system, and finally to give the threshold of social contradiction early warning index (Wang and Huang, 2014). Shi Sen-chang detailed analysis of the traditional social warning and social contradictions early warning difference, not only defines the two objects of study, the real goal, the service object also introduced from the analysis of the difference between the two models (Shi, 2015). Wu Zhong-min defines the concept of social contradictions. He believes that the most appropriate and effective explanation is the meso-level social contradictions. The basic content of social contradiction theory is the social contradiction in the middle level. He discusses the concept of social contradiction and social conflict and social movement (Wu, 2015).

At present, many scholars have designed the evaluation index system of the network public opinion crisis level from different dimensions (Faria et al., 2015). These studies have laid a solid foundation for the research of this paper. However, most of these studies remain in the establishment of indicators and quantification of indicators, the research literature on the early warning of social contradictions is not much, the follow-up application research also needs to be solved.

4. Hypothesis, model and data

4.1. Hypothesis

The comprehensive evaluation index of the previous network public opinion crisis and the current network public opinion crisis evaluation method, we believe that the level of public opinion crisis is determined by the number of netizens who concern about it and the intensity of users' attitudes, that is, "crisis level = number of attentions + attitudes intensity." At the same time, according to "crisis communication and press release," Dr. Shi Anbin thinks that public opinion news and official response are important factors affecting the crisis. We believe that public opinion news, official process response and the official result response as the most important factors affect the trend of public opinion crisis.

Put forward the hypothesis:

1. The level of public opinion crisis caused by public opinion news is explained by the number of netizens who concern about it.

2. The level of public opinion crisis caused by public opinion news is explained by attitude intensity of netizens.

3. The level of crisis caused by official responses to the process is explained by the number of netizens who concern about it.

4. The level of crisis caused by official responses to the process is explained by attitude intensity of netizens.

5. The level of crisis caused by official responses to the result is explained by the number of netizens who concern about it.

6. The level of crisis caused by official responses to the result is explained by attitude intensity of netizens.

4.2. Model

We use the linear model to represent the potential variables of the level of public opinion crisis caused by the public opinion news and public response. The parameter estimation of the multiple regression model is to solve the parameter by the least squares method under the premise that the square of the required error (Ee) is the minimum. The model variables are listed in Table 1.

[Y.sub.1] = [[alpha].sub.1] + [[beta].sub.1][X.sub.1] + [[beta].sub.2][X.sub.2] + [[mu].sub.1] [Y.sub.2] = [[alpha].sub.2] + [[beta].sub.3][X.sub.3] + [[beta].sub.4][X.sub.4] + [[mu].sub.2] [Y.sub.3] = [[alpha].sub.3] + [[beta].sub.5][X.sub.5] + [[beta].sub.6][X.sub.6] + [[mu].sub.3]

In the model, the coefficient [beta]i (i=i ... 6) represents the degree of influence of each independent variable to the crisis level. [alpha] ([alpha]=1, 2, 3) is the intercept term, [mu] ([mu] = 1, 2, 3) is a random error term.

4.3.Seleetion of samples

"2015 China Internet public opinion analysis report" (the report covers the period from January 1, 2015 to October 31, 2015) pointed out that in social contradictions, public management, public security, official anti-corruption and other eight categories. The number of social contradictions hot events was 103, the hot event accounted for up to 20.6%, ranking second. Disputes of social moral, social violence and labor disputes are the three most important social conflicts.

The samples selected in this study are the three typical cases of social conflicts, social violence and labor disputes in the social contradictions from January to October in 2015. The sample data is from the People's Network Public Opinion Monitoring Room coverage period of 500 pieces of public opinion hot events analysis. People's network public opinion monitoring room is one of the earliest professional institutions engaged in Internet public opinion monitoring. It is in the leading position in the field of public opinion monitoring and analysis research, so the research conclusion has certain persuasive power.

4.4.Variables and data

Dependent Variable

The three dependent variables of this study are crisis level Y1 caused by public opinion news, crisis level Y2 caused by official process response, and crisis level Y3 caused by official outcome response. Experts, who work in the People's Network public opinion monitoring room, according to the event development process trigger point: public opinion news and official response, considering the number of users concerned about the attitude of users, using Delphi method to determine the value of the dependent variable. Taking into account the international practice and China's regulatory requirements and the development trend of network public opinion, we use 0-100 points to represent crisis levels. Blue represents the lowest score, the lowest degree of crisis. Red marks the highest score, the highest degree of crisis. As shown in Table 2.

Explanatory Variables and Data

The explanatory variables of this study include two parts: the number of netizens who concern about it and attitude intensity of netizens. Detailed data for each case are shown in table 3.

4.5. Assessment of attitude intensity and the number of netizens Assessment of the number of netizens

We use the public opinion event heat which is invented by people's network public opinion monitoring room to indicate the number of users concerned. Public opinion event heat from newspapers, network news, forums, blogs, micro-blog, WeChat, the six major media keyword search volume weighted and normalized. The weights of the media are as follows: Newspapers: 0.3200; network news: 0.2038; Forum: 0.0752; blog: 0.0954; micro-blog: 0.1409; WeChat: 0.1647. According to the data sources, we believe that public opinion event heat can be expressed as the number of netizens who concern about it.

The experimental evaluation of attitude intensity

We judge the attitude intensity of netizens through the user's comments after public opinion news and official responses to process and result. In this paper, "the network public sentiment text processing and emotion analysis system" is used to analyze the attitude intensity of netizens.

As shown in Figure 1, the system starts from building a corpus, extracting feature words, and then, through the classifier module to classify the text, and finally it will classify the results to the Excel, to achieve visual processing.

The classification method: support vector machine (SVM)

SVM is developed from the optimal classification surface from the linear separable case (Zhang, 2000). SVM can be used to classify the linear separable data. SVM is widely used in the fields of pattern recognition, function estimation, text recognition, time series prediction and so on. Its principle is shown in figure 2. Two different types of samples are represented by solid and hollow points, and the classification line is represented by H. H1 and H2 are the straight lines that are closest to the classification line and are parallel to the classification line, and the distance between them is called margin. The optimal classification line requires that the classification line not only separate the two classes correctly (the training error rate is 0), but also make the classification interval the maximum. The classification line equation is [x.sup.*]w + b=0.

Emotional classification assessment

The evaluation methods are as follows: the average accuracy is (1), positive rate of positive and negative are (2), (3), positive and negative recall rate are (4), (5).

Accuracy = a + b/a + b + c + d Accuracy = a + b/a + b + c + d (1)

Precision (pos) = a/a + b Precision (pos) = a/a + b (2)

Precision (neg) = d/c + d Precision(neg) = d/c + d (3)

Recall (pos) = a/a + c Recall (pos) = a/a + c (4)

Recall (neg) = b/b + d Recall (neg) = b/b + d (5)

The way of data acquisition

In this paper, we select three typical hotspot events in the field of social contradictions. Then, we use the crawler software to collect the netizens online comments about three hot events. These comments come mainly from the comments section after relevant articles in mainstream media related news, post bar, blog, micro-blog and WeChat. In order to ensure the validity of the experimental results, we collect 1000 representative comments for every events of public opinion news, official response to process and result. The collected data is stored in Notepad ++ in text form. The collected comment samples are sorted on the NLCone++. A case study on the attitude intensity of netizens in the case of "illegal vaccine case in Shandong ". The operation is as follows.

a. Locate the initial number to 10.

b. Set the number of categories to 2.

c. Set the feature interval to 50.

d. Output the result data.

In Figure 3, the experimental results show that the Average accuracy =0.83560, which shows that the establishment of the emotional evaluation model accuracy rate is 83%, it can meet the experimental requirements. We have made an experiment on attitude intensity of netizens caused by public opinion news in the case of "illegal vaccine case in Shandong ". In the collection of 1000 comments, the number of comments which the classifier determines positive comments as positive comments is 109, the number of comments which the classifier determines negative comments as negative comments is 891. We believe that the negative comments can reflect the intensity of emotions, so, attitude intensity of netizens = the number of negative comment / total number of comments *100%, attitude intensity of netizens =891/1000*100%=89.1. Subsequently, we use the same method to test these three typical cases, the results shown in the table 5.

5. Model estimation and interpretation

5.1. Model description

We use Statistics SPSS 19 to carry on the multivariate regression test to the dependent variable and the explanatory variable. We conducted three sets of multivariate regression models, (1) the level of crisis caused by public opinion news Y1 as the dependent variable, the number of netizens who concern about public opinion news X1, Attitude intensity of netizens caused by public opinion news X2 as the explanatory variable; (2) the level of crisis caused by process responses Y2 as the dependent variable, the number of netizens who concern about process responses X3, attitude intensity of netizens caused by process responses X4as the explanatory variable; (3) the level of crisis caused by result responses Y3 as the dependent variable, the number of netizens who concern about result responses X5, attitude intensity of netizens caused by result responses X6as the explanatory variable;

5.2. The results of Analysis

In table6 (1), we get the following conclusions by the regression analysis of the level of crisis caused by public opinion news Y1, the number of netizens who concern about public opinion news X1, attitude intensity of netizens caused by public opinion news X2. Adjust [R.sup.2] = 0.998, F =995.080, p=0.000. It can pass the significance test, and it shows that the fitting degree of regression is very high, which can explain the overall 99.8%.

In theory, we think that if the number of netizens who concern about public opinion news and attitude intensity of netizens are zero, the level of public opinion crisis is zero, so the constant term ai does not have practical meaning in the model, the constant term does not affect the validity of the model. The same applies to (2) and (3).

The coefficient of the number of netizens who concern about public opinion news is [beta]1=0.909, the coefficient of attitude intensity of netizens caused by public opinion news is [beta]2=0.22i. This shows that when the other factors remain unchanged, with Xi changes in a unit, the crisis level Y1 changes 0.909 units. Similarly, with X2 changes in a unit, the crisis level Y1 changes 0.221 units. So, hypothesis 1 and hypothesis 2 are correct. Moreover, when the dimension (D) is 3, the condition index (CI) is 78.380, indicating that the data has a serious collinearity. In summary, we believe that the number of netizens who concern about it can be regarded as the only explanatory variables to explain the level of crisis caused by public opinion news.

In table6 (2), we get the following conclusions by the regression analysis of the level of crisis caused by official responses to process Y2, the number of netizens who concern about official responses to process X3, attitude intensity of netizens caused by official responses to process X4. Adjust [R.sup.2]=0.987, F =259.797, p=0.000. It can pass the significance test, and it shows that the fitting degree of regression is very high, which can explain the overall 98.7%.

The coefficient of the number of netizens who concern about official responses to process is [beta]3=0.525, the coefficient of attitude intensity of netizens caused by public opinion news is [beta]4=0.388. This shows that when the other factors remain unchanged, with the X3 changes in a unit, the crisis level Y2 changes 0.525 units. Similarly, with the X4 changes in a unit, the crisis level Y2 changes 0.388 units. That is, hypothesis 3 and hypothesis 4 are correct. Moreover, when the dimension (D) is 3, the condition index (CI) is 35.290, indicating that the data does not have a serious collinearity. In summary, we believe that the level of crisis caused by official responses to the process is explained by both the number of netizens who concern about it and attitude intensity of netizens. Moreover, the number of users concerned about the impact of the degree of 0.525, the impact of attitude intensity of 0.388.

In table6 (3), we get the following conclusions by the regression analysis of the level of crisis caused by official responses to result Y3, the number of netizens who concern about official responses to result X5, attitude intensity of netizens caused by official responses to result X6. Adjust [R.sup.2] =0.985, F =100.106, p=0.000. It can pass the significance test, and it shows that the fitting degree of regression is very high, which can explain the overall 98.5%.

The coefficient of the number of netizens who concern about official responses to result is [beta]5=0.525, but t = 1.354, p> 0.1, it does not pass the significance test, indicating hypothesis 5 is rejected. The coefficient of attitude intensity of netizens caused by official responses to result is [beta]6=0.823, p <0.01. It can pass the significance test. This shows that when the other factors remain unchanged, with the X6changes in a unit, the crisis level Y3 changes 0.823units. That is, hypothesis 6 is correct. Moreover, when the dimension (D) is 3, the condition index (CI) is 65.020, indicating that the data has a serious collinearity. In summary, we believe that to a certain extent, attitude intensity of netizens can be regarded as the only explanatory variables to explain the level of crisis caused by official responses to the result.

6. Conclusions and Recommendations

We analyze the public opinion of social contradictions, and use the regression model to test the hypothesis, the conclusions are as follows:

1. In the initial stage, the number of netizens who concern about it can be regarded as the only explanatory variables to explain the level of crisis caused by public opinion news.

2. At the stage of development, the level of crisis caused by official responses to the process is explained by both the number of netizens who concern about it and attitude intensity of netizens. Moreover, the number of users concerned about the impact of the degree of 0.525, the impact of attitude intensity of 0.388.

3. In the calm phase, attitude intensity of netizens can be regarded as the only explanatory variables to explain the level of crisis caused by official responses to the result.

According to the above characteristics, we propose that in the initial stage, the national government and relevant departments in the network public opinion testing, should focus on the number of netizens who concern about public opinion news. At the stage of development, we should pay attention to both the number of netizens who concern about it and attitude intensity of netizens. In the calm phase, we should focus on attitude intensity of netizens.

Recebido/Submission: 22/07/2016

Aceitacao/Acceptance: 02/10/2016

Acknowledgments

This work was mainly supported by National Social Science Foundation "Research on the warning of social contradictions in the Mongolian and Chinese bilingual base on the big data analysis" (16CGL060); Natural Science Foundation of Inner Mongolia, China (2013MS1009); Subject of Ministry of Education (16XJC710002); Key Scientific Research Project of Inner Mongolia Education Department(NJSZ16026). Inner Mongolia youth talents of science and technology support plan of China (NJYT-15-B08); Inner Mongolia department of soft science project of China "Research on the Key Technologies of Social Governance in Inner Mongolia Network" (2015); Inner Mongolia department of soft science project of China "Research on Contribution of Intangible Assets to Inner Mongolia's Enterprises" (2012); Subject of Inner Mongolia Center for Internet Economic Research.

References

Chen, L, Li, M., Zhang, J., Zeng, W. (2013). Reordering for Chinese-Mongolian SMT Based on Small Parallel Corpus. Journal of Chinese Information Processing, 27(5), 198-204.

Chen, Y., Li, C., Yu, Y., Huang, H. (2012). Threat Evaluation Model of Internet Public Opinion Based on TOPSIS. Journal of Intelligence, 31(3), 15-19.

Dai, Y., Hao, X., Guo, Y., Yu, Z. (2010). Research on the Construction of Network Public Opinion Security Evaluation Index System in China. Netinfo Security, (04), 12-15.

Faria, B. M., Gongalves, J., Reis, L. P., & Rocha, A. (2015). A Clinical Support System Based on Quality of Life Estimation. Journal of Medical Systems, 39(10), 1-11.

Hou, Y. (2012). Word Recognition of Mongolian and Chinese: An EPR Study. Tian Jin: Tianjin Normal University.

Hou, Y., Bai, X. (2012). The Time Course of Phonological Activation and It Role in the Word Recognition: Evidence from the ERP Study. Journal of Inner Mongolia Normal University (Natural Science Edition), 41(1), 51-57.

Shi, G., Zhao, H. (2012). Construction of Social Risk Index of Urban and Rural Areas in China and the Analysis of Group Events. Social Science Research, (1); 68-73.

Shi, S. (2015). Discrimination of early warning of social contradictions. The Journal of Yunnan Administration College, (5), 106-110.

Su, C. (2014). Hierarchical Phrase-based Mongolian-Chinese Statistical Machine Translation. Inner Mongolian University.

Wang, K. Mecklinger, A., Hofinann, J., Weng, X. (2010). From orthography to meaning: anelectro physiological investigation of the role of phonology in accessing meaning of Chinese single-character words. Neuroscience, 165(1), 10 1-106.

Wang, L., Huang, L. (2014). On the Construction of the Early Warning Index System for MacroSocial Contradictions Based on the Effect-to-Cause Reasoning. Journal of shanghai university (social science), 31(5), 79-90.

Wu, Z. (2015). Not All the Contradictions in Society can be Classified as "Social Contradictions": Analysis on the Concept of Social Contradiction. Journal of the Party School of the Central Committee of the C. P. C., 19(2), 51-57.

Zhang W., Qi, J. (2010). Research on the Index System of Public Opinion on Internet for Abnormal Emergency. Journal of Intelligence, 29(11), 71-75.

Zhang X., (2000). Introduction to statistical learning theory and support vector machines. Acta Automatica Sinica, 26(01), 32-42.

Zeng, R. (2010). The Construction of Early Warning Index System of Network Public Opinion. Information Studies: Theory & Application, (01), 77-80.

Zeng, R., Xu, X. (2009). A Study on Early Warning Mechanism and Index for Network Opinion. Journal of Intelligence, 28(11), 52-55.

Jing Yang (1), Yu Ma (2), Linghui Kong (3), Li Yun (4) *

yangj81@126.com_ Yang Jing, 1196321992@qq.com_ Ma Yu, nmgklh@i63.com_Kong Linghui, hyo578@sina.com_ Yun Li

(1) College of Mechanical Engineering in Inner Mongolia University of Technology, Hohhot 010051, China.

(2) College of Management in Inner Mongolia University of Technology, Hohhot 010051, China.

(3) Guangdong University of Finance and Economics, Guangzhou 510320, China.

(4) College of law and Politics in Inner Mongolia Normal University, Hohhot 010051, China.

Table 1--Model variables table

                       Crisis   The number    Attitude
                       level    of netizens   intensity
                                              of netizens

Public opinion news    Y1       X1            X2
Responses to process   Y2       X3            X4
Responses to result    Y3       X5            X6

Table 2--Crisis level representation

Fraction   Crisis level   Color

50-60      I              Blue
61-70      II             Yellow
71-80      III            Orange
81-100     IV             Red

Table 3--Data tables of dependent variables and explanatory
variables

Events             Time and factors        Crisis level   The number
                                                          of netizens

Illegal vaccine    2016.03.18              76             74.9
case in Shandong   Public opinion news
                   2016.03.19              80             77.3
                   Process responses
                   2016.03.21              79             77.9
                   Process responses
                   2016.03.21              45             70.9
                   Result responses
                   8:03:50
                   Result responses        40             75.4
                   19:20:45
8.12 major         2016.08.13              79             78.1
explosion in       Public opinion news
Tianjin port       2016.08.13              89             89.4
                   Process responses
                   2016.08.13-2016.08.15   89             90.9
                   Process responses
                   2016.08.16              88             95.2
                   Process responses
                   2016.08.20              41             69.1
                   Result responses
Lei Yang case      2016.05.09              77             76.9
                   Public opinion news
                   2016.05.09              84             83.3
                   Process responses
                   2016.05.11              88             87.9
                   Process responses
                   2016.05.13              73             74.2
                   Process responses
                   2016.06.30              38             68.1
                   Result responses

Events             Time and factors        Attitude intensity
                                           of netizens

Illegal vaccine    2016.03.18              89.1
case in Shandong   Public opinion news
                   2016.03.19              82.1
                   Process responses
                   2016.03.21              79.4
                   Process responses
                   2016.03.21              44.3
                   Result responses
                   8:03:50
                   Result responses        37.6
                   19:20:45
8.12 major         2016.08.13              90.1
explosion in       Public opinion news
Tianjin port       2016.08.13              88.9
                   Process responses
                   2016.08.13-2016.08.15   89.0
                   Process responses
                   2016.08.16              80.2
                   Process responses
                   2016.08.20              40.3
                   Result responses
Lei Yang case      2016.05.09              87.1
                   Public opinion news
                   2016.05.09              88.7
                   Process responses
                   2016.05.11              89.0
                   Process responses
                   2016.05.13              70.5
                   Process responses
                   2016.06.30              36.3
                   Result responses

Table 4--Experimental results of affective classification

                       Actually positive   Actually negative

Judgment is positive   a                   b
Judgment is negative   c                   d

Table 5--The attitude intensity of netizens table

Events             Time and factors        Positive   Negative
                                           comments   comments

Illegal vaccine    2016.03.18              109        89.1
case in Shandong   Public opinion news
                   2016.03.19              179        82.1
                   Process responses
                   2016.03.21              206        79.4
                   Process responses
                   2016.03.21              557        44.3
                   Result responses
                   8:03:50
                   Result responses        642        37.6
                   19:20:45
8.12 Major         2016.08.13              91         90.1
explosion in       Public opinion news
Tianjin port       2016.08.13              111        88.9
                   Process responses
                   2016.08.13-2016.08.15   110        89.0
                   Process responses
                   2016.08.16              198        80.2
                   Process responses
                   2016.08.20              597        40.3
                   Result responses
Lei Yang case      2016.05.09              129        87.1
                   Public opinion news
                   2016.05.09              113        88.7
                   Process responses
                   2016.05.11              110        89.0
                   Process responses
                   2016.05.13              295        70.5
                   Process responses
                   2016.06.30              637        36.3
                   Result responses

Events             Time and factors        Attitude intensity
                                           of netizens

Illegal vaccine    2016.03.18              89.1
case in Shandong   Public opinion news
                   2016.03.19              82.1
                   Process responses
                   2016.03.21              79.4
                   Process responses
                   2016.03.21              44.3
                   Result responses
                   8:03:50
                   Result responses        37.6
                   19:20:45
8.12 Major         2016.08.13              90.1
explosion in       Public opinion news
Tianjin port       2016.08.13              88.9
                   Process responses
                   2016.08.13-2016.08.15   89.0
                   Process responses
                   2016.08.16              80.2
                   Process responses
                   2016.08.20              40.3
                   Result responses
Lei Yang case      2016.05.09              87.1
                   Public opinion news
                   2016.05.09              88.7
                   Process responses
                   2016.05.11              89.0
                   Process responses
                   2016.05.13              70.5
                   Process responses
                   2016.06.30              36.3
                   Result responses

Table 6--Operation result analysis of linear regression model

                         Regression equation test      Regression
                                                       coefficient

                         Adjust      F         D-W     B
                         [R.sup.2]

1   constant             0.998       995.080   1.152   -11.955
    concerned level                                    0.909
    attitude intensity                                 0.221
2   constant             0.987       259.797   1.956   7.052
    concerned level                                    0.525
    attitude intensity                                 0.388
3   constant             0.985       100.106   2.307   -2.350
    concerned level                                    0.151
    attitude intensity                                 0.823

                         Regression         Collinearity diagnosis
                         coefficient test

                         t                  D   CV      CI

1   constant             -3.872             1   2.997   1.000
    concerned level      42.628             2   0.002   38.271
    attitude intensity   7.039              3   0.000   78.380
2   constant             2.049              1   2.994   1.000
    concerned level      12.279             2   0.003   29.442
    attitude intensity   8.119              3   0.002   35.290
3   constant             -0.454             1   2.995   1.000
    concerned level      1.354              2   0.004   26.248
    attitude intensity   14.016             3   0.001   65.020

Note: D-W: Durbin-Watson; CV: characteristic value; CI: condition
index; D: Dimension
COPYRIGHT 2016 AISTI (Iberian Association for Information Systems and Technologies)
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2016 Gale, Cengage Learning. All rights reserved.

 
Article Details
Printer friendly Cite/link Email Feedback
Author:Yang, Jing; Ma, Yu; Kong, Linghui; Yun, Li
Publication:RISTI (Revista Iberica de Sistemas e Tecnologias de Informacao)
Date:Nov 15, 2016
Words:5297
Previous Article:Research into the influence of CSFs on the performance of PPP projects from a viewpoint of data mining.
Next Article:Survey on e-commerce development in areas inhabited by ethnic minorities of Qujing city, Yunnan province, based on computer data analysis.

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters