دانلود رایگان مقاله انگلیسی وب کاوی برای پیش بینی انتخابات شهردار در تایوان - امرالد 2017

Related work

The practical study of public opinion is not new. Formal ways to measure public opinion are telephone surveys, focus groups, and content analysis. Since survey respondents occasionally provide socially acceptable answers rather than truthfully reporting their behavior, some researchers have devised research strategies that rely on unobtrusive analysis of remarks and speeches to evaluate decision makers’ opinions (Ishiyama and Breuning, 2010). The unobtrusive research method does not require the cooperation of the subject under investigation, and it is relatively low-cost compared with some traditional opinion survey methods. The increasing use of social media in recent years has provided public survey analysts with an opportunity to obtain a large amount of opinions without intruding in the research context. With the help of text mining-based information technologies, analysts can use a set of procedures to make valid inferences from reviews (Ampofo, et al, 2015). For example, Hodge and Matthews (2011) employed text mining to explore the possibilities of incorporating new media data in political environments. The growing need for election forecasting has promoted the generation of much research using weblogs and social media to measure public opinion on elections. For example, Adamic and Glance (2005) analyzed blogs articles with respect to their political orientation, and constructed a network of links among political blogs in 2004 U.S. Presidential election. Albrecht et al.

Conclusion

In Taiwan, most election polls have been conducted using telephone calls for years. With the rise in popularity of the Internet and the rapid development of IT technologies, many people express their opinions about the election on Internet platforms. The major contributions of this study are two-fold. First, in order to address the problem of the Chinese opinions, which are generally expressed in more subtle and complex ways, this study proposes new rules for the extraction of Chinese opinion words associated with attribute words. Second, this study applies Chinese semantic analysis to assist in predicting election results and investigating the topics of concern to voters. To analyze the opinions of the municipal governance-related issues of concern to the citizens in Taichung city, this study classified the collected comments into six topics, including candidates' backgrounds, transport infrastructure, livelihood and public security, arts and culture, social welfare policies, and local economy and industrial construction. Based on the scores of each topic, political parties and candidates can get a clearer idea about the public needs, and utilize this information to plan better policies. We used correspondence analysis to visually present the competition strength of the two mayoral candidates. That the prediction had only 3.62% difference from the real election results shows the effectiveness and accuracy of the proposed prediction method.

This study has some limitations that require further consideration in the future. First, this study focused on revealing Web opinions, and the poll results obtained from traditional method, such as telephone surveys and questionnaires, were not included. To get a more accurate election prediction result, merged methods should be conducted in the future. Second, the opinion database should be updated by considering some newly appeared symbols, such as "囧", an emoticon meaning embarrassment, and "XD", which indicates smiling with eyes narrowed. Third, the extraction rule should capture the ironic sentiments. For example, according to the extraction rule in this study, "○○○候選人好棒棒" would be classified as a positive statement since it includes "好 (very)" and "棒 (excellent)"; However, it is actually a negative statement. Fourth, a sentence that expresses an opinion with a question mark is usually presenting a negative statement or a weak positive statement.