Although Twitter has become a major source of data, research still relies on heavy manual searches to find the accounts of a population of interest, such as members of Parliament, members of an association or candidates to an election. In this article, we propose a method to semi-automatically retrieve those accounts. After creating a pool of possible accounts through automatic web searches, we select a unique account for each member of a defined group. To do so, we use a simplified version of the minimax regret approach, on a score based on the account’s handle and the tweets’ content. We test this method to find the accounts of the candidates for the 2015 Spanish general elections and the members of the 57th House of Commons in the United Kingdom.
Read it here