Node-weighted Graph Convolutional Network for Depression Detection in Transcribed Clinical Interviews DOI
Sergio Burdisso, Esaú Villatoro-Tello, Srikanth Madikeri

et al.

Interspeech 2022, Journal Year: 2023, Volume and Issue: unknown, P. 3617 - 3621

Published: Aug. 14, 2023

We propose a simple approach for weighting selfconnecting edges in Graph Convolutional Network (GCN) and show its impact on depression detection from transcribed clinical interviews.To this end, we use GCN modeling non-consecutive long-distance semantics to classify the transcriptions into depressed or control subjects.The proposed method aims mitigate limiting assumptions of locality equal importance self-connections vs. neighboring nodes GCNs, while preserving attractive features such as low computational cost, data agnostic, interpretability capabilities.We perform an exhaustive evaluation two benchmark datasets.Results that our consistently outperforms vanilla model well previously reported results, achieving F1=0.84 both datasets.Finally, qualitative analysis illustrates capabilities alignment with previous findings psychology.

Language: Английский

AI Methods for Personality Traits Recognition: A Systematic Review DOI

Seyed Mostafa Hashemi Motlagh,

Mohammad Hossein Rezvani,

Mohsen Khounsiavash

et al.

Neurocomputing, Journal Year: 2025, Volume and Issue: unknown, P. 130301 - 130301

Published: April 1, 2025

Language: Английский

Citations

0
Personality Types and Traits—Examining and Leveraging the Relationship between Different Personality Models for Mutual Prediction DOI Creative Commons
Dušan Radisavljević, Rafał Rzepka, Kenji Araki

et al.

Applied Sciences, Journal Year: 2023, Volume and Issue: 13(7), P. 4506 - 4506

Published: April 2, 2023

The popularity of social media services has led to an increase personality-relevant data in online spaces. While the majority people who use these tend express their personality through measures offered by Myers–Briggs Type Indicator (MBTI), another model known as Big Five been a dominant paradigm academic works that deal with research. In this paper, we seek bridge gap between MBTI, and Enneagram Personality, goal increasing amount resources for model. We further explore relationship was previously reported MBTI types certain traits well test presence similar measures. propose new method relying on psycholingusitc features selected based This approach showed best performance our experiments up 3% automatic recognition per-trait level. Our detailed experimentation offers insight into nature how it translates different models.

Language: Английский

Citations

10
Tools, Potential, and Pitfalls of Social Media Screening: Social Profiling in the Era of AI-Assisted Recruiting DOI
Yeqing Kong, Huiling Ding

Journal of Business and Technical Communication, Journal Year: 2023, Volume and Issue: 38(1), P. 33 - 65

Published: Sept. 18, 2023

Employers are increasingly turning to innovative artificial intelligence recruiting technologies evaluate candidates’ online presence and make hiring decisions. Such social media screening, or profiling, is an emerging approach assessing influence, personalities, workplace behaviors through their publicly shared data on networking sites. This article introduces the processes, benefits, risks of profiling in employment decision making. The authors provide important guidance for job applicants, technical professional communication instructors, professionals how strategically respond opportunities challenges automated technologies.

Language: Английский

Citations

10
Can ChatGPT read who you are? DOI Creative Commons
Erik Derner, Dalibor Kučera, Nuria Oliver

et al.

Computers in Human Behavior Artificial Humans, Journal Year: 2024, Volume and Issue: 2(2), P. 100088 - 100088

Published: July 26, 2024

The interplay between artificial intelligence (AI) and psychology, particularly in personality assessment, represents an important emerging area of research. Accurate trait estimation is crucial not only for enhancing personalization human-computer interaction but also a wide variety applications ranging from mental health to education. This paper analyzes the capability generic chatbot, ChatGPT, effectively infer traits short texts. We report results comprehensive user study featuring texts written Czech by representative population sample 155 participants. Their self-assessments based on Big Five Inventory (BFI) questionnaire serve as ground truth. compare estimations made ChatGPT against those human raters ChatGPT's competitive performance inferring text. uncover 'positivity bias' assessments across all dimensions explore impact prompt composition accuracy. work contributes understanding AI capabilities psychological highlighting both potential limitations using large language models inference. Our research underscores importance responsible development, considering ethical implications such privacy, consent, autonomy, bias applications.

Language: Английский

Citations

2
TranSentGAT: A Sentiment-Based Lexical Psycholinguistic Graph Attention Network for Personality Prediction DOI Creative Commons
Shahryar Salmani Bajestani, Mohammad Mahdi Khalilzadeh, Mahdi Azarnoosh

et al.

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 59630 - 59642

Published: Jan. 1, 2024

Extended use of social media, lead to personality detection from online content shared by users. While it has numerous applications in different areas such as recommendation systems, most existing studies focus on superficial, statistical, and explicit user contents, ignoring the knowledge hidden semantic features. In this study, we proposed a method explore psycholinguistic deep levels users data for task prediction. We utilizing fine-tuned domain-specific BERT model extract features sentence, enriched outputs leveraging emotional information highlight important words. Furthermore, conducting double-way-attention mechanism reflected highlighted words into whole extracted inputs. Then, created graph considering embeddings last step node developing dynamic task-realted learning approach specify edges connect pairs nodes based neural network, leveraged attention network predict traits. Finally, experimental results confirmed effectiveness our with 80.63% accuracy, compared other state-of-the-art essays dataset. Also, several ablations are conducted illustrate verify impact sections parameteres architecture.

Language: Английский

Citations

1
Personality trait analysis during the COVID-19 pandemic: a comparative study on social media DOI Creative Commons
Marcos Fernández-Pichel, Mario Ezra Aragón,

Julián Saborido-Patiño

et al.

Journal of Intelligent Information Systems, Journal Year: 2023, Volume and Issue: 62(1), P. 117 - 142

Published: Aug. 28, 2023

Abstract The COVID-19 pandemic, a global contagion of coronavirus infection caused by Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), has triggered severe social and economic disruption around the world provoked changes in people’s behavior. Given extreme societal impact COVID-19, it becomes crucial to understand emotional response people on personality traits psychological dimensions. In this study, we contribute goal thoroughly analyzing evolution aspects large-scale collection tweets extracted during pandemic. objectives research are: i) provide evidence that helps estimated pandemic temperament, ii) find associations trends between specific events (e.g., stages harsh confinement) reactions, iii) study multiple aspects, such as degree introversion or level neuroticism. We also examine development emotions, natural complement automatic analysis To achieve our goals, have created two large collections (geotagged United States Spain, respectively), collected Our work reveals interesting dimensions, events. For example, period, found increasing traces Another insight from is most frequent signs disorders are those related depression, schizophrenia, narcissism. some peaks negative/positive emotions

Language: Английский

Citations

3
Can ChatGPT Read Who You Are? DOI Creative Commons
Erik Derner, Dalibor Kučera, Nuria Oliver

et al.

arXiv (Cornell University), Journal Year: 2023, Volume and Issue: unknown

Published: Jan. 1, 2023

The interplay between artificial intelligence (AI) and psychology, particularly in personality assessment, represents an important emerging area of research. Accurate trait estimation is crucial not only for enhancing personalization human-computer interaction but also a wide variety applications ranging from mental health to education. This paper analyzes the capability generic chatbot, ChatGPT, effectively infer traits short texts. We report results comprehensive user study featuring texts written Czech by representative population sample 155 participants. Their self-assessments based on Big Five Inventory (BFI) questionnaire serve as ground truth. compare estimations made ChatGPT against those human raters ChatGPT's competitive performance inferring text. uncover 'positivity bias' assessments across all dimensions explore impact prompt composition accuracy. work contributes understanding AI capabilities psychological highlighting both potential limitations using large language models inference. Our research underscores importance responsible development, considering ethical implications such privacy, consent, autonomy, bias applications.

Language: Английский

Citations

2
Node-weighted Graph Convolutional Network for Depression Detection in Transcribed Clinical Interviews DOI
Sergio Burdisso, Esaú Villatoro-Tello, Srikanth Madikeri

et al.

Interspeech 2022, Journal Year: 2023, Volume and Issue: unknown, P. 3617 - 3621

Published: Aug. 14, 2023

We propose a simple approach for weighting selfconnecting edges in Graph Convolutional Network (GCN) and show its impact on depression detection from transcribed clinical interviews.To this end, we use GCN modeling non-consecutive long-distance semantics to classify the transcriptions into depressed or control subjects.The proposed method aims mitigate limiting assumptions of locality equal importance self-connections vs. neighboring nodes GCNs, while preserving attractive features such as low computational cost, data agnostic, interpretability capabilities.We perform an exhaustive evaluation two benchmark datasets.Results that our consistently outperforms vanilla model well previously reported results, achieving F1=0.84 both datasets.Finally, qualitative analysis illustrates capabilities alignment with previous findings psychology.

Language: Английский

Citations

2