Emoji Extractions from Geotagged Twitter Data
We provide a filtered, pre-processed and anonymized dataset collected originally from the Twitter decahose (a random 10% sample of Twitter) over 29 days in October, 2016, to support computational social science research on how people on Twitter use emojis. The data comprises of a table with four columns and 4,057,872 rows (including a header). The fields of the table are: ID: A unique tweet ID that could also be used to ‘hydrate’ the contents of the tweet directly from Twitter. Country: The country code associated with the tweet. Language: The language metadata tag associated with the tweet. We only retain the top 30 languages (sorted by frequency of tweets) in our initial corpus. Emojis: The emojis extracted from the ‘text’ field of the tweet. The methodology that was used to extract the tweets is described in the next section. A paper that uses the data as the basis for its findings and also contains descriptive statistics: M. Kejriwal, Q. Wang, H. Li, L. Wang, An Empirical Study of Emoji Usage on Twitter in Linguistic and National Contexts. Online Social Networks and Media. In Press.