OpenStreetMap and Hadoop

I need some ideas for a weekend project about Hadoop and OpenStreetMap.

I have access to AWS EC2 instance with OpenStreetMap snapshot in my EBS volume. The OpenStreetMap data is in a PostgreSQL database.

What kind of MapReduce function can be run on the OpenStreetMap data, assumin开发者_开发问答g I can export them into xml format, and then place into HDFS ?

In other words, I am having a brain cramp at the moment, and cannot think what kind of MapReduce operation that can extract valuable insight from the OpenStreetMap xml?

(i.e. extract all the places designated as park or golf course. But this needs to be done once only, not continuously)

Many Thanks

• An interesting idea of my coworkers is to find nodes which are not connected, but should be. For example a street which abruptly ends and continues 1m further in a separate path.
• Precalculate shortest paths