New software accurately predicts personality based on tweets

London: In some good news for employers, researchers are developing a software that can predict a job candidate’s personality and behaviour by analysing their postings on Twitter. People who care for the welfare of others frequently use the words ‘we’, ‘us’, ‘friends’, ‘family’, ‘health’ and ‘home’ on the micro-blogging site, according to an analysis by IBM research in San Jose, California. [caption id=“attachment_1246733” align=“alignleft” width=“380”]  Twitter. Reuters image[/caption] Those who do not have those qualities are more likely to use the words ‘work’, ‘school’, ‘job’, ‘music’ and ‘beer’. Perfectionists can be identified by their use of the words ‘perfect’ and ‘chauffeur’. People who settled for just good enough used the words ‘mix’, ‘half’ and ‘fix’. The researchers said 80 per cent of the results were highly correlated with the findings of in-depth psychometric surveys on the same 300 people who had their Twitter profiles assessed by the software, ‘The Times’ reported. Researchers have interviewed 4,243 employees to collect 450,000 pieces of data to ensure that the analysis measures the sort of characteristics sought by employers. They work on the premise that five basic psychological traits are needed for the right balance: extroversion, being agreeable, conscientious and open, and having the correct dash of neuroticism. The software scans a target’s Twitter postings to assess their mix. People who are energetic and warm use longer words and make more sexual references in their tweets; those who are altruistic and likely to fit in with their colleagues use more first person plurals, more indefinite and definite articles and swear less often, the report said. Conscientious people discuss their achievements and use language to highlight distinctions while the worriers use words about being unwell, discuss negative emotions and talk less about topics such as family, it said. Using shorthand such as ‘ur’ or ‘2day’ is a sign of a lack of openness. PTI