WDRA Extender

Sample Extended Data from a Twitter Search about Corona Virus Ventilators

The WDRA Extender service uses the Twitter API to fill out data captured from the WDRA (Web Data Research Assistant) Chrome extension.

The WDRA browser plugin allows researchers to capture Twitter data without any programming expertise. Its key value is in being able to capture data from historic searches, which is impossible with the Twitter API and any services built on the API. The weakness of WebDataRA is that it is limited to data that is visible in the Web page. This complementary web service allows you to backfill the WebDataRA spreadsheet with extra information obtained from the Twitter API or with analysis that is difficult to produce in a browser. It is anticipated that more powerful text and network analysis will be provided to facilitate more sophisticated computation social science research methodologies.

Instructions

Simply paste a list of Tweet IDs from your spreadsheet (a column of text with each line starting 'ID:' and followed by 14 digits) into the box below. Press the "Submit" button, and the following files will be returned to you in a ZIP package:

• spreadsheet.csv - an extended version of the normal Web Data RA spreadsheet, with extra data in the account table (e.g. name, profile description, location, number of followers, date of registration). Selected information (e.g. number of followers) is also inserted in the tweet table, to allow the tweets to be sorted by properties of their authors (e.g. to determine which tweets are associated with thge most significant actors who have the highest number of followers). The spreadsheet also includes a new table (next to the hashtag table) of the most common words (a frequency table omitting those words that occur only once).
• wordcloud.png - a wordcloud based on the santised text column, allowing you to see a visula representation of the most popular words/topics. Word clouds may be a rather niaive data visualisation, but they can offer an important instant visual overview.
• kwik.html - a keyword in context table of the tweets (taken from the sanitised text column) allows you to see how important (or frequent) words and phrases are used within the tweets, to dismbiguate the different senses of each word and to provide an alternative overview of the content of the tweets.
• network.html - a visualisation of the conversational network formed when one account "replies to" or "mentions" another account. (Note that rewteets are not taken into account.) Nodes are sized according to the number of the account's followers, and coloured according to the number of tweets that they have contributed to the conversation (on a spectrum from light yellow to dark red indicating few to many tweets). Hovering over a node will give the full name and profile of the account, as well as the precise number of tweets and followers.

NB: you may paste partial lists of IDs from "cleansed" and "sampled" sets of Tweets that you have previously collected. You may combine tweet ids from multiple sets of searches, and from other sources. This service will create a new collection and recalculate all the statistics from scratch. Your list may be composed of thousands of tweet ids. This service uses TWARC to obtain tweet and user data from the Twitter API, but it is subject to Twitter's rate limiting and so it is possible that a very large request may take several hours to satisfy.

Disclaimer: THIS SOFTWARE IS NOT YET AVAILABLE OUTSIDE THE AUTHOR'S OFFICE. Even when installed, this version will be a pilot service that provides no queue management in the first instance. A new version is being developed by the University of Southampton Research Software Group.