Tsunami, News, and Time

The tsunami born out of the massive 9.0 earthquake in the Indian Ocean took two hours to reach Sri Lanka, but there was no warning.  When the earthquake occured, another tsunami was born: tsunami of information.  Unprepared, tsunami of seawater across the ocean outraced tsunami of information across fabric of technology over social networks.

We can build a tsunami warning system in little time with just what we already have: networks of telephones, instant messaging, e-mail, news feeds, TV, and radio.  If spammers can 'inform' millions of people with a single button, any one of us could have sent tsunami warnings to all the TV and radio stations in the Indian Ocean.

This tragedy reminded me how important time really is in news processing.  While Google News obviously doing much more than time analysis, you can get a lot with simple keyword search for news articles created within a range of time that starts with approximate time of event.  Also general culture or region specific shape of news propagation can be used to increase or decrease search weights.

By combining time analysis with language analysis, it should be possible to identify smaller waves of news within a larger wave.  In a sense, each propation of news is a synchronization in language, meaning we tend to use the words we are exposed to.

Implementation is rather simple but time consuming.  You look for a set of keywords and time range that returns the desired shape.  This process can be short-circuited by increasing weights on words used by search engine users during the same time period.

Hmm.  Looks like I got side-tracked.