FAQ


What is "PoliTwi USA"?

PoliTwi USA consists of POLItics and TWItter. It's an extension of PoliTwi for German politics and its aim is to analyze freely available tweets on political issues in the United States of America. Here, the hashtags concerning the political businesses are analyzed to identify in particular current trends.

Who is behind the project?

It's a research project at the Institute for Information Systems (iisys) at Hof University in collaboration with the Big Data Analytics Research Lab at Goethe University Frankfurt. Besides PoliTwi USA we deal with the automatic analysis of textual data (text mining), especially with the the extraction of opinions (opinion mining).

How does "PoliTwi USA" work?

Based on well-chosen search terms (GOP, Democrats, Obama, Senate, tcot, ccot, pjnet, ...) we collect political tweets according to politics in the USA in Twitter. We anaylyze these tweets to detect current political trends.

Thus, spying my data?

NO! We only analyze what everyone can read in Twitter. We don't collect user-related data.

What's the meaning of the numbers and evaluations?

The results shown aren't a scientific established analysis and the data isn't representative in any way. However they give a general mood of the Twitter community. In particular, emerging topics and discussions can be detected early. New topics appear shortly after their distribution on Twitter.

But what's the sense?

PoliTwi detects trends in Twitter almost immediately. If there are many tweets written according to a new topic, this is listed in the section "Topics of the previous hour" very fast.

And what exactly is the technical story?

We programmed a framework in Java, which consists of a streaming-component, one PostgreSQL database, different analysers, modules for topic-detection and statistical analysis and many other modules. To illustrate the historical data we use DataTables designed and created by Allan Jardine.

What are topic-points?

We compute the rank of every topic regarding on how "new" or "relevant" each topic is. For this purpose, the tweets are counted per hour or per day and set in relation to a certain previous period.

How prevailing is the data?

The data and the findings of our analysis are updated continuously. The counter for "total tweets" will be displayed in real time. The top topics of the previous hour are always calculated 5 minutes past full hour. The topics of the previous day are available form 0:15 AM EST (Eastern Standard Time). If there are no topics for one period, it is pointed out with a yellow sign on the corresponding list. The top topics for the previous period are displayed instead. This problem occurs at night and early in the morning, because there is lower activity on Twitter and fewer tweets are written (people should sleep at night :-)).

Why is Twitter used?

The social network Twitter is a very popular service in the social web. In Twitter new topics will be discussed very early. Especially in the USA Twitter is the most occupied social service, 64 percent of the whole population is registered.

What does the number at "total tweets" mean?

This value shows the total number of analyzed tweets in realtime

What are topics?

Topics are the latest political issues of the previous hour respectively the last day.

How many tweets are analyzed?

For PoliTwi USA the analysis started at October 22, 2014 at 4:58 PM EST (Eastern Standard Time). Each day, we analyze about 400,000 tweets. On Senate Election in November 2014 we expect about twice as much tweets particularly.

Are there any questions about PoliTwi?

You can contact us at politwi@iisys.de :-)