Human values can be defined as abstract long-lasting features serving as principles that guide a person's life. They are fundamental for understanding a series of socio-psychological phenomena, such as environmental attitudes and behaviors, religiosity, prejudices, drug use, antisocial behavior, juvenile delinquency, suicidal tendencies and pro-environment attitudes.
The functional theory of human values identifies two consensual value functions. Th efirst function has three types of value orientations: social, central and personal, while the second has two types: humanitary and materialistic. As shown in the figure below, the product of these two functions creates six value subfunctions (normative, realization, existence, suprapersonal, interactive and experimentation).
Following this theory, we created a self-report inventory to measure the level of importance attributed by people to certain values. The questionnaire consists of 18 items, which are answered using a Likert scale going from 1 (totally not important) to 7 (totally important).
This work aims to create and evaluate models for human value classification. For that, we analyze the viability of using machine learning to predict results from the human value questionnaire using messages taken from Twitter. With such a classifier, it would be possible to define a person's human values based only on messages shared by them, without needing any extra activities. Thus making it easier to comprehend and explain social and psychological phenomena.
This project resulted in a paper titled Human Values Classification in Social Network Using Machine Learning, which can be found at: https://sol.sbc.org.br/index.php/eniac/article/download/9270/9172.