WorldTrends was a project to track keywords and activity in news sources and trends in financial markets to correlate positive and negative world events with news and financial trends. Future work was to try to extract some pattern or predictive algorithm. This project collected data faithfully for about year before requiring more server storage and computation capacity than I had available at the time, and was put on hold.
WorldTrends 1.0 captured headlines and numerical indices for several mass media and niche news sources, as well as financial market data and headlines. It provided a web-based viewing platform for browsing this data, and also offered realtime tools to examine the data in cloud tagging views, charts, and graphs. WorldTrends 1.0 was based on a Web 1.0 model - server-generated images produced by PHP, with page refreshing to update data as it is collected by the server. In addition, I began WorldTrends 2.0, built on the Web 2.0 paradigm, to AJAXify and modernize the WorldTrends experience.
In a Cemetech news article announcing WorldTrends 1.0, I mentioned some interesting results in its first two months of data (about topics suited to early 2007):
- WorldTrends recorded 2,133,121 instances of ‘iPhone’ yet only 1,812,815 mentions of ‘Vista’.
- ‘Iraq’ appears 6,761,922 times, while ‘Bush’ appeared a comparable 5,238,684 times with much the same peak distribution.
- ‘Democrat’ showed up 1,550,929 times, while ‘Republican’ was only mentioned 440,659 times.
- In a staggering imbalance, ‘war’ showed up 4,708,975 times, while a mere 386,911 instances of ‘peace’ were uncovered.