Figure - uploaded by Ryan L. Boyd
Content may be subject to copyright.
Raw visual features where there is a significant difference (p < 0.05) between male and female images. Positive effect sizes indicate that women prefer the feature, while negative effect sizes indicate that men prefer the feature.
Source publication
Background / Introduction. This work explores the relationship between a person's demographic/psychological traits (e.g., gender, personality) and self-identity images and captions. Methods. We use a dataset of images and captions provided by N ≈ 1, 350 individuals, and we automatically extract features from both the images and captions. Results. W...
Context in source publication
Context 1
... C, E, A, and N stand for Openness, Conscientiousness, Extraversion, Agreeableness, and Neuroticism, respectively. Table 3 shows effect sizes for features significantly different between men and women. As suggested by previous research, men are more likely to use the color black [56]; other correlations appear to confirm stereotypes, e.g., a stronger preference by women for pink and purple. ...Similar publications
As the big data and internet growing huge day by day, there is in need to overwhelm people by the information in large. These kind of issues makes researchers to create a technique that can generate automatic text. This research is very great interesting topic and the summarization of text is the process which undergoes finding important informatio...
Citations
... The idea that language use reveals information about personality has long circulated in the social and medical sciences. Previous research has demonstrated that the way people use words convey a great deal of information about themselves and their mental health conditions [1][2][3][4], including academic success [5]; however, much of the previous research has focused on the analysis of self-reports or essays. In contrast, implicit motives, which are indicators used by professional psychologists during the aptitude diagnosis, are not readily accessible to the conscious mind and, therefore, not detected using self-reports of personal needs, or through essay writing [6]. ...
... To perform our experiments, we employed the dataset available in the GermEval 2020 shared task on the "Classification and Regression of Cognitive and Motivational style from the text". 1 The provided data, mostly written in standard German language, has been collected from around 14000 subjects that participated in the OMT test. 4 Each answer was manually labeled with the motives (0, A, L, M, F) and the levels (from 0 to 5). ...
According to the psychological literature, implicit motives allow for the characterization of behavior, subsequent success, and long-term development. Contrary to personality traits, implicit motives are often deemed to be rather stable personality characteristics. Normally, implicit motives are obtained by Operant Motives, unconscious intrinsic desires measured by the Operant Motive Test (OMT). The OMT test requires participants to write freely descriptions associated with a set of provided images and questions. In this work, we explore different recent machine learning techniques and various text representation techniques for facing the problem of the OMT classification task. We focused on advanced language representations (e.g, BERT, XLM, and DistilBERT) and deep Supervised Autoencoders for solving the OMT task.
We performed an exhaustive analysis and compared their performance against fully connected neural networks and traditional support vector classifiers. Our comparative study highlights the importance of BERT which outperforms the traditional machine learning techniques by a relative improvement of 7.9%. In addition, we performed an analysis of how the BERT attention mechanism is being modified. Our findings indicate that the writing style features acquire higher importance at the moment of accurately identifying the different OMT categories. This is the first time that a study to determine the performance of different transformer-based architectures in the OMT task is performed. Similarly, our work propose, for the first time, using deep supervised autoencoders in the OMT classification task. Our experiments demonstrate that transformer-based methods exhibit the best empirical results, obtaining a relative improvement of 7.9% over the competitive baseline suggested as part of the GermEval 2020 challenge. Additionally, we show that features associated with the writing style are more important than content-based words. Some of these findings show strong connections to previously reported behavioral research on the implicit psychometrics theory.
... Recently, researchers focused on studying how the personality traits are manifested on social networks. Such researches proved the presence of relationships between the Big Five traits and traditional features extracted from users" generated data (i.e., text and images) [6][7][8]. On the same side, many psychological studies managed to examine the relationships between the personality traits and some psychological characteristics. These characteristics have proven to be good predictors for many personality traits. ...
... These studies made use of all possible features which can be extracted from an individual"s personal profile. To enhance the assessment"s performance, researchers focused on studying the relationships between personality traits and different social networks related features [7,[11][12][13]. Most of these studies focused on applying text analysis to extract traditional textual features such as words count, part of speech (POS) tags, and N-grams [14]. ...
Recently, researches focused on studying how the Big Five personality traits are manifested on social networks. These researches proved the presence of relationships between the Big Five Personality traits and various social networks features extracted from users' generated content. In this paper, the relationships between the Big Five personality traits (Openness to experience, Conscientiousness, Extraversion, Agreeableness, and Neuroticism) and attributes of personality characteristics identified as the Personal Values and Human Needs are studied. These attributes or namely features, are extracted from users' posts on social media. The relationship between the traits and proposed attributes is investigated through Pearson correlation coefficients. A dataset for 564 Twitter users is used in an experimental study, where findings proved the presence of relevant correlations between the traits and the proposed personality characteristic features. The Conscientiousness, Agreeableness, and Neuroticism traits showed strong relations existence with all of the Personal Values features, while the Openness to experience and Neuroticism traits showed strong correlations with Liberty and Self-expression Needs features consecutively. The proposed study verified the effectiveness of the proposed Personal Values and Human Needs features as indicators for the Big Five personality traits, proving their ability for personality characteristics classification.
... Moreover, sentiments and emotions are complex cognitive processes, with rich features that are difficult to interpret and portray through a single modality alone (Qiu, Liu, & Lu, 2018). Hence, multimodal methods for SER gained increasing interest (Zhao et al., 2019;Kumar, Srinivasan, Cheng, & Zomaya, 2020), especially in light of recent studies which showed that they can substantially improve the prediction of emotional states by overcoming the limitations of unimodal systems (Burdick, Mihalcea, Boyd, & Pennebaker, 2021). ...
Prior approaches for multimodal sentiment and emotion recognition (SER) exploit input data representations and neural networks based on the classical Euclidean geometry. Recently, however, the hyperbolic metric proved to be a powerful tool for data mapping, being able to capture the hierarchical structure of the relations among elements in the data. In this paper we propose the use of hyperbolic learning for SER, and show that the inclusion in the neural network of hyperbolic structures mapping the input into the hyperbolic space can improve the quality of the predictions. The benefits brought by the hyperbolic features are evaluated by developing extensions of existing methods following two approaches. From one side, we modified state-of-the-art models by including hyperbolic output layers. From the other, we generated hybrid neural network architectures by combining hyperbolic and Euclidean layers according to different schemes. The proposed hyperbolic models were tested on several classification tasks applied to benchmark multimodal SER datasets. Experiments gave strong evidence that in both simple and complex networks the introduction of a hyperbolic structure results in an improvement of the model accuracy. Specifically, the combined use of hyperbolic and Euclidean layers showed superior performance in almost all the classification tasks.
... For future work, we plan to improve the current techniques for tweets authorship profiling, such as gender [17], age group [59,60]. Alternative deep learning networks with early exits [61], and ensemble-based approaches [62] will also be explored. ...
Stance detection is a relatively new concept in data mining that aims to assign a stance label (favor, against, or none) to a social media post towards a specific predetermined target. These targets may not be referred to in the post, and may not be the target of opinion in the post. In this paper, we propose a novel enhanced method for identifying the writer's stance of a given tweet. This comprises a three-phase process for stance detection: (a) tweets preprocessing; here we clean and normalize tweets (e.g., remove stop-words) to generate words and stems lists, (b) features generation; in this step, we create and fuse two dictionaries for generating features vector, and lastly (c) classification; all the instances of the features are classified based on the list of targets. Our innovative feature selection proposes fusion of two ranked lists (top-k) of term frequency-inverse document frequency (tf-idf) scores and the sentiment information. We evaluate our method using six different classifiers: K nearest neighbor (K-NN), discernibility-based K-NN, weighted K-NN, class-based K-NN, exemplar-based K-NN, and Support Vector Machines. Furthermore, we investigate the use of Principal Component Analysis and study its effect on performance. The model is evaluated on the benchmark dataset (SemEval-2016 task 6), and the results significance is determined using t-test. We achieve our best performance of macro F-score (averaged across all topics) of 76.45% using the weighted K-NN classifier. This tops the current state-of-the-art score of 74.44% on the same dataset.
... The shifting landscape of human experience has made the subtleties of behavioural traces particularly important: the physical and digital footprints that human behaviour leaves behind are a veritable goldmine of personality data (Lambiotte & Kosinski, 2014). The images that a person chooses to share with others (Burdick, Mihalcea, Boyd, & Pennebaker, 2020;, a person's words and other features of their verbal behaviour Golbeck, 2016;Hoover, Dehghani, Johnson, Iliev, & Graham, 2018;Kern et al., 2014;Mitra, Counts, & Pennebaker, 2016;Park et al., 2015), URL clicks (Lien, Bai, & Chen, 2019;Tellakat, Boyd, & Pennebaker, 2019), social behaviours (Adali & Golbeck, 2012;Hilbig, Thielmann, Hepp, Klein, & Zettler, 2015), and self-presentation behaviours (Liu, Preotiuc-Pietro, Samani, Moghaddam, & Ungar, 2016;Segalin et al., 2017;Shiramizu, Kozma, DeBruine, & Jones, 2019;Todorov et al., 2005;Walker, Schönborn, Greifeneder, & Vetter, 2018) are no longer simply reflections of personality-they are the critical raw material for understanding personality itself. ...
Personality psychology has long been grounded in data typologies, particularly in the delineation of behavioural, life outcome, informant-report, and self-report sources of data from one another. Such data typologies
are becoming obsolete in the face of new methods, technologies, and data philosophies. In this article, we discuss personality psychology’s historical thinking about data, modern data theory’s place in personality psychology, and several qualities of big data that urge a rethinking of personality itself. We call for a move away from self-report
questionnaires and a reprioritization of the study of behaviour within personality science. With big data and behavioural assessment, we have the potential to witness the confluence of situated, seamlessly interacting psychological
processes, forming an inclusive, dynamic, multiangle view of personality. However, big behavioural data come hand
in hand with important ethical considerations, and our emerging ability to create a ‘personality panopticon’ requires
careful and thoughtful navigation. For our research to improve and thrive in partnership with new technologies, we
must not only wield our new tools thoughtfully, but humanely. Through discourse and collaboration with other disciplines and the general public, we can foster mutual growth and ensure that humanity’s burgeoning technological capabilities serve, rather than control, the public interest.
The Linguistic Inquiry and Word Count (LIWC) is a popular closed-vocabulary text analysis software program that is used to understand whether individuals’ use of linguistic categories (i.e., word categories, such as negative affect) depends on their personality traits. Here, we present the first meta-analysis of the relations between the Big Five personality traits and 52 linguistic categories of the English language. Across 31 eligible samples (n = 85,724), the results showed that (a) self-reported personality traits are significantly correlated with linguistic categories, but the effect sizes are relatively small (the strongest effect sizes between the Big Five and linguistic categories ranged from |ρ| = .08 to .14, and the 52 LIWC categories explained on average 5.1% of personality variance); (b) observer-reported personality traits are significantly correlated with linguistic categories, with the effect sizes being small-to-medium (|ρ| = .18–.39, explaining 38.5% of personality variance); (c) 20 linguistic categories (out of 260; 5 Personality Traits × 52 LIWC Categories) correlated both with self- and observer-reported personality traits (the “kernel of truth” in linguistic markers of personality); and (d) 10 study, sample, and task characteristics significantly moderated the correlations of the linguistic categories with personality traits, showing that the effect sizes were mainly stronger for longer texts and older LIWC versions, among others.
Past literature has shown that extraversion is related to the use of positive emotion and social process words. However, the strength of the relationships varies substantially across studies. In this research, we conducted a meta-analysis (k = 37, N = 82,132) to estimate the overall effect size of the two linguistic correlates of extraversion. In addition, we tested potential moderators including demographic variables (e.g., age and gender) and communication contexts (e.g., synchronous vs. asynchronous, public vs. private). Our random effects models revealed a small correlation between extraversion and positive emotion words (r = .069, 95% CI = [.041, .096]), and a small correlation between extraversion and social process words (r = .077, 95% CI = [.044, .109]). In addition, the strength of the relationship between extraversion and positive emotion words varies across communication contexts, while the relationship between extraversion and social process words remains consistent across contexts. Our results suggest that positive emotion words and social process words are linguistic correlates of extraversion, but they are small in magnitude.
Past literature has shown that extraversion is related to the use of positive emotion and social process words. However, the strength of the relationships varies substantially across studies. In this research, we conducted a meta-analysis (k = 37, N = 82,132) to estimate the overall effect size of the two linguistic correlates of extraversion. In addition, we tested potential moderators including demographic variables (e.g., age and gender) and communication contexts (e.g., synchronous vs. asynchronous, public vs. private). Our random effects models revealed a small correlation between extraversion and positive emotion words (r = .069, 95% CI = [.041, .096]), and a small correlation between extraversion and social process words (r = .077, 95% CI = [.044, .109]). In addition, the strength of the relationship between extraversion and positive emotion words varies across communication contexts, while the relationship between extraversion and social process words remains consistent across contexts. Our results suggest that positive emotion words and social process words are linguistic correlates of extraversion, but they are small in magnitude.
