Dataset Open Access

TweetsCOV19 - A Semantically Annotated Corpus of Tweets About the COVID-19 Pandemic (Part 2, May 2020)

Baran, Erdal; Dimitrov, Dimitar

Citation Style Language JSON Export

  "publisher": "Zenodo", 
  "DOI": "10.5281/zenodo.4593502", 
  "author": [
      "family": "Baran, Erdal"
      "family": "Dimitrov, Dimitar"
  "issued": {
    "date-parts": [
  "abstract": "<p><strong><a href=\"\">TweetsCOV19</a></strong><strong> </strong>is a semantically annotated corpus of Tweets about the COVID-19 pandemic. It is a subset of <a href=\"\">TweetsKB</a> and aims at capturing online discourse about various aspects of the pandemic and its societal impact. <strong>Metadata</strong> information about the tweets as well as extracted <strong>entities</strong>, <strong>sentiments</strong>, <strong>hashtags</strong>, <strong>user mentions</strong>, and <strong>resolved URLs </strong>are exposed in RDF using established RDF/S vocabularies*.</p>\n\n<p>We also provide a <em><strong>tab-separated values (tsv)</strong></em> version of the dataset. Each line contains features of a tweet instance. Features are separated by tab character (&quot;\\t&quot;). The following list indicate the feature indices:</p>\n\n<ol>\n\t<li>Tweet Id: Long.</li>\n\t<li>Username: String. Encrypted for privacy issues*.</li>\n\t<li>Timestamp: Format ( &quot;EEE MMM dd HH:mm:ss Z yyyy&quot; ).</li>\n\t<li>#Followers: Integer.</li>\n\t<li>#Friends: Integer.</li>\n\t<li>#Retweets: Integer.</li>\n\t<li>#Favorites: Integer.</li>\n\t<li>Entities: String. For each entity, we aggregated the original text, the annotated entity and the produced score from <a href=\"\">FEL</a> library. Each entity is separated from another entity by char &quot;;&quot;. Also, each entity is separated by char &quot;:&quot; in order to store &quot;original_text:annotated_entity:score;&quot;. If FEL did not find any entities, we have stored &quot;null;&quot;.</li>\n\t<li>Sentiment: String. <a href=\"\">SentiStrength</a> produces a score for positive (1 to 5) and negative (-1 to -5) sentiment. We splitted these two numbers by whitespace char &quot; &quot;. Positive sentiment was stored first and then negative sentiment (i.e. &quot;2 -1&quot;).</li>\n\t<li>Mentions: String. If the tweet contains mentions, we remove the char &quot;@&quot; and concatenate the mentions with whitespace char &quot; &quot;. If no mentions appear, we have stored &quot;null;&quot;.</li>\n\t<li>Hashtags: String. If the tweet contains hashtags, we remove the char &quot;#&quot; and concatenate the hashtags with whitespace char &quot; &quot;. If no hashtags appear, we have stored &quot;null;&quot;.</li>\n\t<li>URLs: String: If the tweet contains URLs, we concatenate the URLs using &quot;:-: &quot;. If no URLs appear, we have stored &quot;null;&quot;</li>\n</ol>\n\n<p>To extract the dataset from <a href=\"\">TweetsKB</a>, we compiled a seed list of 268 COVID-19-related <a href=\"\">keywords</a>.</p>\n\n<p><em>* For the sake of privacy, we anonymize&nbsp;user IDs&nbsp;and we do not provide the text of the tweets.</em></p>", 
  "title": "TweetsCOV19 - A Semantically Annotated Corpus of Tweets About the COVID-19 Pandemic (Part 2, May 2020)", 
  "type": "dataset", 
  "id": "4593502"
All versions This version
Views 1,0781,078
Downloads 353353
Data volume 88.2 GB88.2 GB
Unique views 1,0281,028
Unique downloads 273273


Cite as