Construction of MBTI Personality Estimation Model Considering Emotional Information

Construction of MBTI Personality Estimation Model Considering Emotional Information Ryota Kishima, Kazuyuki Matsumoto, Minoru Yoshida, Kenji Kita Division of Science and Technology, Graduate School of Sciences and Technology for Innovation, Tokushima University, Japan c612135039@tokushima-u.ac.jp {matumoto;mino;kita}@is.tokushima-u.ac.jp Abstract The widespread use of social networking ser- vices (SNSs) such as Twitter and Facebook in recent years have rendered it easy to in- teract with various people all over the world. Analysis of the user personality and emotion using features such as the text information and image information on SNSs has been re- searched. This study analyzes the effective- ness of using emotional information as a fea- ture when building a personality estimation model for Twitter users. The extensively used Myers-Briggs type indicator (MBTI) is used for analysis. A method is proposed for esti- mating the MBTI personality pattern by ex- tracting emotional information from Twitter and adding it to the characteristics of the dis- tributed expressions of sentences and words. Through evaluation experiments, it is deter- mined whether the emotional information is effective in constructing the MBTI personal- ity estimation model. The obtained results demonstrate that the accuracy is improved when emotional information is included as a feature, compared to the case without emo- tional information. 1 Introduction In recent years, it has become possible for anyone to easily interact with many others all over the world because of the prevalence of social networking ser- vices (SNSs) such as Twitter and Facebook. Cer- tain services recommend suitable friend candidates to the user based on information such as hobbies, occupation, age, and gender. However, it is difficult to comprehend the personality of the user based on this information alone. Therefore, we investigate the estimation of the user personality based on the com- ments posted on SNSs. Bakry et al. (Bakry et al., 2019) described per- sonality estimation based on user information from online social networks (OSNs). They not only sum- marized several machine learning algorithms, such as the SVM, Naive Bayes, gradient boosting, and neural networks, but also described the features and dataset scale used for training, considering each character estimation model. Some of the models have a high accuracy rate of approximately 90% but as the data set is small, there is no generalization. The other model accuracies are not high, at approxi- mately 60%. Thus, the accuracy of personality es- timation based on user information from SNSs is insufficient. However, it is crucial to improve the personality estimation accuracy for the realization of various services on SNSs. Yamada et al. (Yamada et al., 2019) extracted features from user tweets and built a personality estimation model using a support vector machine (SVM). They utilized the Myers-Briggs type indi- cator (MBTI), which is an introspective self-report questionnaire. As shown in Table 1, it includes four indicators with two types each, and a total of 16 per- sonality types. In our research, we not only use the sentence vector extracted from a tweet sentence but also the emotional information estimated from it as a feature …

Content truncated to 500 words