- 上一篇：袜子分拣机英文文献和中文翻译
- 下一篇：VB编程英文文献和翻译
2.3. Web Content Mining
Web mining is an application of data mining techniques to discover patterns in the web. According to analysis subjects, web mining can be pided into three different types; web usage mining, web content mining and web structure mining[2,4,11,16]. Web usage mining is the process to extract useful information from server logs i.e. user history. Web usage mining is the process to find out what users are looking for on the Internet. Web structure mining uses graph theory to analyze nodes and connection structure of a web site. According to the type of web structure data, Web structure mining can be pided into two kinds. The first type is to extract patterns from hyperlinks in the web. The second type is to analyze document structure of page structures to describe HTML or XML tags. Web content mining aims to discovery useful information from the web contents, data, and documents. Recently, many researchers focused on opinion mining as a part of text mining. Opinion mining tries to extract and analyze the means of consumer’s opinions on the specific products in the Internet. The texts used in opinion mining can be from opinions stored including comments in blogs, consumers’ opinions in news sites [3,4,14].
3. Research Method
Because this study aims to analyze the relationship between buzz share and market share, the correlation between buzz share and market share of movie ticketing are investigated. The detail experimental phases are as follows.
Step1: The subject movies are selected as top-two movies per every week, which data is gathered from movie box-office of Movie Promotion Council in Korea.
Step2: For top-two movies at time t, the buzz share of t-2, t-2, t, t+1, and t+2 are gathered from blog, café, Jisik-in (Q&A), News, video services of Naver, a representative Internet portal site in Korea. The numbers of references of movies are translated to the ratios of buzz share of movies.
Step 3: In the third phase, the relationships of buzz share and market share are tested using correlation
Step 4: In the fourth phase, we try to interpret the results of the experiment. analysis.
This study conducts correlation analysis between two variables defined as weekend ticket sales (market share) and the buzz share. The results of the correlation analysis depending in time lags on Tab.1 and Fig.1.
4. Result
Tab.1 The correlation between ticket reservation and buzz share of various channels depending on time[8]
The analysis results can be summarized as follows.
The buzz shares can work as leading indeces of market share since the buzz share in t-1 time point is positively correlated with market share.
The effects of leading indexes are diminishing as time period is far from current time. That is, the correlations of t-2 time point with market share at t time point are lower than those of t-1 time point.
The buzz share is also lagging indexes of market share. This can be interpreted from the fact that the buzz shares of t+1 and t+2 points are highly correlated with market shares.
The `Video' is a more significant leading index because at t-2 time point, only the correlation coefficient of `Video' is significantly as 0.232.
The correlation coefficients with buzz shares from `Video' and `News' are higher than those of other channels at a point of t-1. This can be happen because `Video' or `News' are usually made film distributing agencies in order to promote their films.
At the point of t+1 and t+2, consumer-oriented buzz shares such as blog and community are higher correlated with actual movie ticket reservation than commercial channels such as `Video' and`News'.
5. Conclusion
This study examines the correlation between buzz shares on the Internet and the actual market share of movies tickets. The empirical experiment results show that there exist correlations between buzz shares and actual ticket sales. Also, the investigation results show that different channels of the Internet have different leading and lagging effects on actual market share. Especially the correlations between buzz shares in video sites and news sites, and ticket sales are higher at t-1 and t-2 time points than blog and Internet community, which means that buzz shares in video sites and news sites are more valuable as leading indexes of market share. However, the correlations between buzz shares in blog and Internet community, and actual ticket sales are higher at t+1 and t+2 points, which means that buzz shares in blog and Internet community are more valuable as lagging indexes of market share.