Haruno Masahiko, Principal Investigator of the Center for Information and Neural Networks (SINET) and Drs. Mori Kazuma, National Institute of Information and Communications Technology (NICT, President: TOKUDA Hideyuki, Ph.D), reports the use of machine learning. Analyze behavior on Twitter and predict a wide range of personality traits and characteristics such as intelligence and excess.
Specifically, the study uses component-wise gradient boosting to demonstrate network features, such as the number of tweets and the number of likes, and the use of the term on Twitter as social (eg, wastage) and mental health. Predictions of (eg, anxiety) are personality respectively. This approach may provide a new way to mental health diagnosis and personal nudity.
The new study was published online in the Journal of Personality on Thursday August 20, 2020.
Social media services (SNS) have quickly become universal tools for communication. Previous research has shown that information about the use of Facebook and Twitter can reveal basic and course personality traits based on the Big 5.. However, the extent to which SNS information can be used to indicate specific personality traits and characteristics is unknown. Personality traits and characteristics are being predicted by analyzing SNS information and how accurately the personality represents the user.
Dr. The study by MORI and theory investigator HARUNO found that an analysis of the behavior of four different types of users on Twitter (ie, network features, time, word stats, and word usage) showed a wide range of personality traits and characteristics. Can be estimated.
A statistical analysis found significant correlations between measured personality and trait scores and perceived estimates, with correlation coefficients around 0.25. This value is not sufficient to accurately determine an individual’s personality traits, but with an adequate population sample, this technique can provide informative results.
The study collected social media information from 239 participants (156 men, 83 women; mean age 22.4 years) who also underwent personality tests in which 24 personality traits and characteristics (52 subscales) were measured. Of the 52 sub-centers, Twitter information can be used reliably to predict 23 of them.
The study shows a positive correlation (correlation coefficient = 0.44) between measured and predicted Big 5 extroversion scores based on a 10-fold cross-validation procedure performed 10 times (Bonferroni corrected a P value of 0.05 / 52).
The analysis revealed that many social personalities such as extraversion, empathy, and autism can be estimated from network features. Other personality traits such as socioeconomic status, smoking / drinking and even depression or schizophrenia were inferred from language use features.
Time prediction was more difficult to make correlations with measured personalities, but showed a significant correlation with intelligence and social value orientation.
We are expanding the analysis to thousands of topics. The method described in this study can be used to act on people’s behaviors for mental health diagnosis and individual behavior. It will also give insight on the nervous system underlying individual differences in personality traits.
Social media services (SNS) have quickly become universal tools for communication. Previous research has shown that information about the use of Facebook and Twitter can reveal basic and course personality traits based on the Big 5.
However, which types of SNS information can be used to indicate specific personality traits and characteristics are unknown. Personality traits and characteristics are being predicted by analyzing SNS information and how accurately the personality represents the user.
Dr. The study by MORI and theory investigator HARUNO found that a wide range of personality traits and characteristics can be estimated by analyzing the behavior of four different types of users on Twitter (ie, network features, time, words Statistics and word usage).
A statistical analysis found significant correlations between measured personality and trait scores and perceived estimates, with correlation coefficients around 0.25. This value is not sufficient to accurately determine an individual’s personality traits, but with a large enough population sample, this technique can provide informative results.
The study collected social media information from 239 participants (156 men, 83 women; mean age 22.4 years) who also underwent personality tests in which 24 personality traits and characteristics (52 subscales) were measured. Of the 52 sub-centers, Twitter information can be used reliably to predict 23 of them.