Initial Filters

Foursquare

  1. Find relevant context in the Tweets

Workflow

  1. Construct Twitterer Objects from the main collection of Tweets (Post filters)
  2. Analyze each Twitterer Object: 1. Run DBScan clustering algorithm on each User.

Determining Location during the event:

  1. Find user clusters. Do not limit the timespan anymore, simply find user clusters.
  2. With each cluster, examine the following aspects: a. Tweeting Consistency? b. Holes in time for that cluster?

Note: All clusters will have temporal holes in them. What is important is to find the biggest clusters, since the dataset is so much bigger now.

Determining Tweet Consistency

Using the same time blocking method as before, a T_Score is assigned to a cluster, this is now defined as 

time blocks / number_of_tweets**2