Dynamic Collocational Network of Plasticizer News 塑化劑新聞動態網絡

I've been working on visualizing collocational networks for a while.  As demonstrated earlier (here), a collocational network of Japan earthquake had been  visualized from simply news texts.  This time, the goal is to make collocational network a dynamic network.  That is, the network changes over time.


Since a social event is never static, news reports of this event , as well, change over time.  In this sense, time itself is a critical factor that influences the construction of collocational network.  The basic idea is to break down time-span.  My previous demonstration showed a static network across a large time-span (a month).  Now, I simply break it down into small pieces (a day), and re-connect these pieces by time sequence (date).  And you can see how this idea works in the video at the top of this page.

In the video, I tried to visualize the Plasticizer scandal in Taiwan.  If you were aware of Taiwan's situation, the Plasticizer scandal is really a big news during the time.  I collected all available news at Yahoo! Taiwan every day from May 23 to June 23, 2011.  Each day is a time frame.  And, each collocational network is built in the basis of a single day news.  With the help of Gephi, these separated networks are combined to a dynamic network.

For all frames, the nuclear word is 塑化劑(Plasticizer).  So the word 塑化劑(Plasticizer) stays unchanged.  But, core words are also fixed, such as 食品(food), 產品(product), 果汁(juice), 檢驗(exam), etc.  However, peripheral words are blinking, and the connections sometimes just disappear.  In this context, we can easily grasp an outlook of the event, even if the event is changing. 

As a contrast, a static collocational network of Plasticizer scandal is listed below.