Twitter in Webometric Analyst

To search for and download tweets, please use Mozdeh rather than Webometric Analyst. There are some additional functions in Webometric Analyst that can analyse the files downloaded by Mozdeh (see below) just in case you need them.

Running a basic time series analysis of tweets

See http://mozdeh.wlv.ac.uk/installation.html

Making networks from Twitter

See more detailed Twitter conversation network and Twitter key term/user list identification instructions.

To make a network of Twitter Users based on their friend and/or follower relationships, first create a plain text file with one user name per line, each starting with @. Then submit this file to the Make Network From Users' Connections button. This takes a long time – perhaps only 5 users can be processed per hour. The results will be used to build a network of Friend and/or Follower connections.

To make a network of Twitter Users based on their Tweets, where a connection A->B exists for each tweet sent by A that includes @B, start by running searches for tweets (either a single search or with the monitoring option). Then send the result to the Create network from @user connections in tweets button to create a network of the top tweeters. Choose the number of tweeters to include in the network – more than 500 will make the network slow to draw and more than 5000 may cause memory problems and crash the program.

More details:

Creating networks of following/followers in Twitter

Three types of following/follower network can be created from lists of users:

  1. Following networks
  2. Follower networks
  3. Following and follower networks

Following/follower lists can be extracted from Twitter using Webometric Analyst for a specified list of users. These should typically be key actors for your topic. The process is slow if they have >1000 followers/following because of the Twitter rate limits. In any case, it is only possible to identify the first 5000 followers/following.

Method
  1. Using Windows Notepad, create a plain text file with a list of usernames, one per line.
  2. Start Webometric Analyst, close down the start-up Wizard and click the Twitter tab.
  3. Click the List Twitter Following/Followers button and select your plain text file of Twitter usernames and follow the instructions. You will be asked to login to Twitter for permission.
  4. Wait for the program to finish. If your list has many users or they have many friends or followers then this may take a long time.
  5. When the program has finished, convert the list into a network with a second Webometric Analyst button, Make Network From following/follower Connections File.
When the network appears, tidy it by moving the nodes so that the labels do not overlap and/or colouring the key nodes yellow.

 

See also the section: Analysing the timelines of tweeters.

Please see also: Thelwall, M. (2015). Evaluating the comprehensiveness of Twitter Search API results: A four step method. Cybermetrics,18-19, p1.