Matching Theory and Data with Personal-ITY - What a Corpus of Italian YouTube Comments Reveals About Personality

Sep 17, 2026

Uploaded by Asher

As a contribution to personality detection in languages other than English, this work builds Personal-ITY, a new corpus of Italian YouTube comments labeled with MBTI personality traits via distant supervision. Personality prediction experiments evaluate performance on Personal-ITY and on TWISTY, a Twitter dataset also annotated with MBTI. An in-depth feature analysis links model signals back to the psychological theory underlying the labels, showing no single best model and varying trait detectability and theory alignment.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages12

Reading Time30 min

AccessFree to download

Abstract

1 Introduction

2 Background

2.1 Psychological models

Matching Theory and Data with Personal-ITY: What a Corpus of Italian YouTube Comments Reveals About Personality Elisa Bassignana}~ Malvina Nissim| Viviana Patti~~Dipartimento di Informatica, Universit degli Studi di Torino, Italy }Dpartment of Computer Science, IT University of Copenhagen, Denmark CLCG – Faculty of Arts, University of Groningen, The Netherlands ~ viviana .p ti@unito.it, }[elba@itu.dk](elba@itu.dk)[m.nissim@rug.nl](m.nissim@rug.nl) Abstract As a contribution to personality detection in languages other than English, we rely on distant supervision to create Personal-ITY, a novel corpus of YouTube comments in Italian, where authors are labelled with personality traits. The traits are derived from one of the mainstream personality theories in psychology research, named MBTI. Using personality prediction experiments, we (i) study the task of personality prediction in itself on our corpus as well as on TWISTY, a Twitter dataset also annotated with MBTI labels; (ii) carry out an extensive, in-depth analysis of the features used by the classiﬁer, and view them speciﬁcally under the light of the original theory that we used to create the corpus in the ﬁrst place. We observe that no single model is best at personality detection, and that while some traits are easier than others to detect, and also to match back to theory, for other, less frequent traits the picture is much more blurred. 1 Introduction Human Personality is a psychological construct aimed at explaining the wide variety of human behaviours in terms of a few, stable and measurable individual characteristics (Snyder, 1983; Parks and Guay, 2009; Vinciarelli and Mohammadi, 2014) . Research in psychology has formalised these characteristics into what are known as Trait Models. Two are the major models widely adopted also outside of purely psychological research (see Section 2.1): Big Five (John and Srivastava, 1999) and Myers-Briggs Type Indicator (MBTI) (Myers and Myers, 1995) . The psychological tests commonly used to detect prevalence of traits include human judgements regarding semantic similarity and relations between adjectives that people use to describe themselves and others. This is because language is believed to be a prime carrier of personality traits (Schwartz et al., 2013) . This aspect, together with the progressive increase of available user-generated data from social media, has prompted the task of Personality Detection, i.e., the automatic prediction of personality from written texts (Whelan and Davies, 2006; Argamon et al., 2009; Celli et al., 2013; Youyou et al., 2015; Litvinova et al., 2016; Verhoeven et al., 2016) . Personality detection can be useful in predicting life outcomes such as substance use, political attitudes and physical health. Other ﬁelds of application are marketing, politics, psychological and social assessment and, in the computational domain, dialogue systems (Ma et al., 2020) and chatbots (Qian et al., 2018) . As a contribution to personality detection in languages other than English, we have developed Personal-ITY, a novel corpus of YouTube comments in Italian, which are annotated with MBTI personality traits. The corpus creation methodology, described in detail in (Bassignana et al., 2020), makes use of a Distant Supervision approach that can also serve as a blueprint to develop datasets for otherlanguages. In this work, we use Personal-ITY to cast light not only on the feasibility of personality detection per se on this corpus, but also on the relationship between our data and the theory it is based on. More speciﬁcally, we want to investigate if our distantly obtained labels are meaningful with respect to the psychological theory they come from, and whether language does indeed reﬂect the traits that should be associated with such labels. To do this, we run a series of in-and cross-dataset experiments; on top of This work is licensed under a Creative Commons Attribution 4 .0 International Licens

FAQs

What is Personal-ITY and how is it created?

Personal-ITY is a corpus of Italian YouTube comments labeled with MBTI personality traits. It is built using a distant supervision approach.

What datasets and experiments are used for personality prediction?

Experiments evaluate personality prediction on Personal-ITY and also on TWISTY, a Twitter dataset annotated with MBTI labels.

How do the results relate machine-learning features to psychological theory?

The work conducts an extensive feature analysis and interprets classifier signals in light of the MBTI-based theory used to create the corpus, finding no single best model and varying trait detectability.

Sep 17, 2026

Uploaded by Asher