Supervised sentiment and topic classification of Twitter/X data on cultural heritage in the post-earthquake recovery of L’Aquila, Italy (tweet and sentence level)

<p dir="ltr">This dataset contains the manually (supervised) classified subset of tweets used to study the role of cultural heritage in the recovery of L’Aquila (Italy) ten years after the 6 April 2009 earthquake.</p><p dir="ltr">The source corpus is 4,349 tweets posted between 4 and 10 April 2019 with the hashtag #L’Aquila, purchased from a third-party vendor. Two subsets are published here:</p><p dir="ltr">1.Tweet level (14 records): tweets in which cultural heritage is the main topic (0.3% of the corpus).

Fields: record number, date, tweet text (translated into English), language, polarity (positive / negative / neutral), the share of sentences per class (Pt = positive, Ng = negative, Nt = neutral, Unr = unrelated), links included in the tweet, and the text extracted from those links.</p><p dir="ltr">2.Sentence level (243 records): tweets in which cultural heritage is one of the topics (5.6% of the corpus). Same fields, plus the dominant topic of the tweet (20 categories, e.g. reconstruction, restoration, commemoration actions, urban facilities, solidarity messages, cultural heritage, seismic information, governance).</p><p dir="ltr">Polarity and topics were assigned by the authors using a set of predefined classification rules, cross-checked between authors and moderated by the first author.