A Survey of Automatic Personality Detection from Texts

Proceedings of the 28th International Conference on Computational Linguistics, pages 6284–6295 Barcelona, Spain (Online), December 8-13, 2020 6284 A Survey of Automatic Personality Detection from Texts Sanja ˇStajner and Seren Yenikent∗ Symanto Research Nuremberg, Germany {sanja.stajner, seren.yenikent}@symanto.com Abstract Personality profiling has long been used in psychology to predict life outcomes. Recently, au- tomatic detection of personality traits from written messages has gained significant attention in computational linguistics and natural language processing communities, due to its applicability in various fields. In this survey, we show the trajectory of research towards automatic personality detection from purely psychology approaches, through psycholinguistics, to the recent purely natural language processing approaches on large datasets automatically extracted from social media. We point out what has been gained and what lost during that trajectory, and show what can be realistic expectations in the field. 1 Introduction Personality is a collection of different constructs such as thoughts, feelings, and values which underlie individual differences and predict human behavior (Roberts and Mroczek, 2008). Due to its complex and multifaceted structure, automatic detection of personality requires a holistic understanding of the construct, which is not an easy task even considering today’s technological advancements. Throughout the personality research history, attempts for personality modelling ranged from tradi- tional psychology methods (e.g. questionnaires), via psycholinguistic approaches (e.g. counting specific word types in texts), to the recent purely natural language processing (NLP) approaches that attempt at detecting personality traits from large amounts of social media data. Especially with recent technolog- ical advancements and big data, the research area has extended with the premise that digital footprints could capture not only indirect and natural characteristics but also psychological insights on deeper lev- els. However, the research has shown a number of problems raising from such approaches which we systematically discuss in this survey. In Section 2, we introduce the two widely-used personality models, and discuss their similarities and dissimilarities in details. In Section 3, we present use cases of personality detection in the fields of com- putational linguistics (CL), natural language processing (NLP), and artificial intelligence (AI). Section 4 shows the trajectory of automatic personality detection, from the early pure psychology approaches to the latest NLP approaches on large social media datasets, pointing out strengths and weaknesses of each of the approaches. Section 5 further discusses computational problems raising from using Twitter data for the Myers-Briggs Type Indicator (MBTI) personality modelling, while Section 6 discusses ethical con- cerns regarding automatic personality detection in general. In Section 7, we revisit the main conclusions of the presented survey. 2 Personality Models The most widely and frequently used personality models are The Big 5 Model (Costa and McCrae, 1992) and the MBTI model (Briggs-Myers and Myers, 1995). Both of them are long-established psychological models that have attracted attention from CL and NLP fields in hope to offer their wider usage in industry and everyday life. ∗* Both authors have contributed equally to this work. This work is licensed under a Creative Commons Attribution 4.0 International License. License details: …

Content truncated to 500 words