Abstract:

To determine whether and how city halls in Poland use Twitter. An attempt to identify the information that is sent through tweets. Identification of city hall profiles on Twitter. Acquisition of tweets' URLs and their content. The text mining analysis of the tweets included text pre-processing, creation of corpora of the documents, creation of a document-term matrix, and application of classical methods deriving from data mining. Information on the number of users (cities hall) and the number of tweets published by them. The most frequently used hashtags. The number of tweets published per hour and day of the week. The number of hashtags, users mentioned, and links in tweets. Hidden, abstract topics describing tweets generated with the LDA algorithm.

Profiles on Twitter were identified using the Google search engine. Due to the number of cities, no requests to city halls concerning their Twitter profiles were sent. Only the content of the tweets and the time they were sent were analyzed. Organizations can use the applied method to compare their Twitter activity with other organizations of the same type. The study found that text mining of tweet content helps determine which groups of information are provided by city halls. It was also determined whether they included links to external sources or hashtags in their tweets and whether they mentioned other users.