Development of a Japanese Personality Dictionary based on Psychological Methods

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 3103–3108 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 3103 Development of a Japanese Personality Dictionary based on Psychological Methods Ritsuko Iwai, Daisuke Kawahara, Takatsune Kumada, Sadao Kurohashi Kyoto University1, RIKEN BSI-TOYOTA Collaboration Center2 1Yoshida-honmachi, Sakyo-ku, Kyoto-shi, Kyoto, JAPAN, 2Hirosawa, Wako-shi, Saitama, JAPAN iwai.ritsuko.8n@kyoto-u.ac.jp {dk, t.kumada, kuro}@i.kyoto-u.ac.jp Abstract We propose a new approach for constructing a personality dictionary with psychological evidence. In this study, we collect personality words by using word embeddings and construct a personality dictionary with weights for the Big Five traits. The weights are calculated based on the responses of the large sample (N = 1,938, female = 1,004, M = 49.8 years old: 20–78, SD = 16.3). All the respondents answered a 20-item personality questionnaire and 537 personality items derived from word embeddings. We present the procedures to examine the qualities of responses with psychological methods and to calculate the weights. Using these, we develop a personality dictionary with two sub-dictionaries. We also discuss an application of the acquired resources. Keywords: personality, psychology, weights 1. Introduction When describing ourselves or others, we usually use certain abstract terms such as “she is a sociable person” or “he is kind.” With such cues, we expect that the person is friendly or warm. We can also imagine that they accept an invitation to a party or help their classmates without hesitation. These assumptions can be made because social knowledge about latent traits are shared among humans. This is critical in human social perceptions, and even unconsciously affects human behavior (Ferguson and Bargh, 2004). It allows them to infer others’ personalities based on the words that describe their personalities. For understanding human personality based on the trait descriptions of others, computers need knowledge on the relationships between personality descriptors and personality traits. However, to the best of our knowledge, such a personality dictionary does not exist, which is applicable to natural language processing (NLP). In this study, we develop a Japanese personality dictionary with weights for Big Five traits, incorporating psychological methods. Moreover, our approach is the first NLP study that has theoretical and statistical evidence tolerant to psychological research standards. The Big Five is one of the most widely used framework to understand universal personalities (e.g., McCrae and Costa, 1997). The Big Five assumes that human individual differences in personalities can be described in five broad traits: Extraversion, Agreeableness, Conscientiousness, Neuroticism, and Openness-to- Experiences1 (Goldberg, 1992). EX indicates the degree to which a person is extraverted; AG indicates that for agreeable, warmth, and sympathetic; CO for self- disciplined, organized, and motivated; NE for sensitivity, worrisome, and anxious; OP for the aspects of curiosity and intelligence (Gosling et al., 2003). We calculate the weights of personality words per the Big Five traits to infer a person’s personality based on the personality word they are described by. With such weights, researchers can 1 In this paper, Extraversion is hereinafter referred to as EX, …

Content truncated to 500 words