search_tweets function

perform full text search on tweets collected with epitweetr

perform full text search on tweets collected with epitweetr (tweets migrated from epitweetr v<1.0.x are also included)

search_tweets(
  query = NULL,
  topic = NULL,
  from = NULL,
  to = NULL,
  countries = NULL,
  mentioning = NULL,
  users = NULL,
  hide_users = FALSE,
  action = NULL,
  max = 100,
  by_relevance = FALSE
)

Arguments

• query: character. The query to be used if a text it will match the tweet text. To see how to match particular fields please see details, default:NULL
• topic: character, Vector of topics to include on the search default:NULL
• from: an object which can be converted to ‘"POSIXlt"’ only tweets posted after or on this date will be included, default:NULL
• to: an object which can be converted to ‘"POSIXlt"’ only tweets posted before or on this date will be included, default:NULL
• countries: character or numeric, the position or name of epitweetr regions to be included on the query, default:NULL
• mentioning: character, limit the search to the tweets mentioning the given users, default:NULL
• users: character, limit the search to the tweets created by the provided users, default:NULL
• hide_users: logical, whether to hide user names on output replacing them by the USER keyword, default:FALSE
• action: character, an action to be performed on the search results respecting the max parameter. Possible values are 'delete' or 'anonymise' , default:NULL
• max: integer, maximum number of tweets to be included on the search, default:100
• by_relevance: logical, whether to sort the results by relevance of the matching query or by indexing order, default:FALSE If not provided the system will try to reuse the existing one from last session call of setup_config or use the EPI_HOME environment variable, default: NA

Returns

a data frame containing the tweets matching the selected filters, the data frame contains the following columns: linked_user_location, linked_user_name, linked_user_description, screen_name, created_date, is_geo_located, user_location_loc, is_retweet, text, text_loc, user_id, hash, user_description, linked_lang, linked_screen_name, user_location, totalCount, created_at, topic_tweet_id, topic, lang, user_name, linked_text, tweet_id, linked_text_loc, hashtags, user_description_loc

Details

epitweetr translate the query provided by all parameters into a single query that will be executed on tweet indexes which are weekly indexes. The q parameter should respect the syntax of the Lucene classic parser https://lucene.apache.org/core/8_5_0/queryparser/org/apache/lucene/queryparser/classic/QueryParser.html

So other than the provided parameters, multi field queries are supported by using the syntax field_name:value1;value2 AND, OR and -(for excluding terms) are supported on q parameter. Order by week is always applied before relevance so even if you provide by_relevance = TRUE all of the matching tweets of the first week will be returned first

Examples

if(FALSE){
   #Running the detect loop
   library(epitweetr)
   message('Please choose the epitweetr data directory')
   setup_config(file.choose())
   df <- search_tweets(
        q = "vaccination", 
        topic="COVID-19", 
        countries=c("Chile", "Australia", "France"), 
        from = Sys.Date(), 
        to = Sys.Date()
   )
   df$text
}

See Also

search_loop

detect_loop