Class Projects - Spring 2013

twXplorer

TwXplorer allows journalists and casual users explore Twitter’s search results more deeply and quickly identify interesting tweets and trends.

TwXplorer project from the Collaborative Innovation class Spring 2013.
What it does

For journalists and news readers searching on Twitter, it's often difficult to get deep in to the results and surface interesting, not widely-disseminated information. TwXplorer returns users a set of tweets that match their search criteria, but it goes a step further by charting the most frequently used terms. Users can select from terms in the chart to drill-down into the results and identify interesting subsets of tweets. They can also exclude words from their results, view previous search snapshots to see how topics have changed over time, and export their results into a CSV file for other uses.

The tool gives both journalists and casual users the power to explore Twitter’s search results more quickly and thoroughly and provides them with a set of features enabling them to identify interesting tweets about a certain topic. Journalists can use twXplorer to search for story topics and identify popular retweets, find existing articles that have been written about the topic, and save tweets that interest them.


How it works

TwXplorer queries Twitter for tweets matching a user's search terms and stores the results in a database. The system then queries the database for all results matching the search term and parses through the text of the tweets to generate the lists of the most frequently used word in those tweets. The database is accessed again with a different query if a user drills down in to the most frequently used words. At each level of analysis the system finds the most frequently used words and generates a graph to help users visualize the results. When the user selects the export feature, a CSV file is generated with the all the results from the database matching the search term. See the architecture diagram for the application for more detail.


Next Steps
  • Search within specific date ranges
  • The ability to export tweets to Storify
  • Currently the number of tweets returned for a search is limited. Future iterations would be more comprehensive. 

Results of a search for "NYC bike sharing."

Connect

Live demo from student project

Code repository

twXplorer became a Knight Lab prototype. Check it out.


Initial Concept: Larry Birnbaum and Rich Gordon

Student Team: Allen Zeng (computer science), Jeanette Huang (computer science), Miguel Huerta (journalism)

Faculty Guidance: Larry Birnbaum, Rich Gordon, and Kris Hammond (with assistance from Shawn O’Banion)