ameerkat / esp

Natural disaster detection and prediction using social media

About

Note 2/9/2012 - Pretty much on permanent hiatus. ESP is a project designed to detect natural disasters from social media sources (specifically twitter), make predictions of affected areas and provide alerts to people living in those areas automatically.

Setup

To run you must create a Options.cs file of the following format, the stream url is just the base url to the stream you are using, the methods used for searching tend to rely on the filter.json stream method to provide keyword and geo based searching.

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;

namespace Tweet
{
	class Options
	{
		public static String twitterStreamUrl = "https://stream.twitter.com/1/statuses/filter.json";
		public static String twitterUsername = your_twitter_username;
		public static String twitterPassword = your_twitter_password;
		public static int numberThreads = 4;
	}
}

TODO

• Add in code to push deserialized tweets into a database
• Add in a routine to do event detection and push events to another database
  • We can use dbscan to do this http://www.c-sharpcorner.com/uploadfile/b942f9/implementing-the-dbscan-algorithm-using-C-Sharp/
  • http://en.wikipedia.org/wiki/DBSCAN
• Add some web services (could event be simple json over http) to access the events so that the client (mobile app or demo webpage etc.) can get event data, the web services should just interface the end client with our tweet/event database.
• Consolidate TwitterProducer and TwitterConsumer into a class inherited from "StatusProducer" which will allow for multiple services (Facebook, Twitter, Google+) to be used and produce a uniform "Status" object.
• Use Silverlight/Bing Maps API To plot tweets: http://stackoverflow.com/questions/2162626/silverlight-add-pushpin-to-bing-maps-via-c-sharp