University of Sheffield
Automatic Personality Recognition from Non-verbal Acoustic Cues: Bridging the Gap Between Psychology and Computer Science
Abstract
dc:description.abstractHuman-computer interaction (HCI) is an evolving research field; it has changed from focusing on usability and interface design to more complex interaction and adaptivity design. Many disciplines have become involved in HCI including psychology. One interesting aspect of psychology, which is relevant to HCI is personality. Personality is a stable pattern of behaviour and thought that uniquely characterizes individuals and how they behave in social contexts. There are many personality theories but the big five, introduced in 1980 by Lewis Goldberg, is the most successful and widely used. The big five are openness, conscientiousness, extraversion, agreeableness, and neuroticism, known as OCEAN. In the last ten years, there has been a growing interest in `social signals' since its introduction by Alex Pentland in 2007. Social signals were later defined in 2010 by Poggi and D'Errico as ``communicative or informative signal that, either directly or indirectly, conveys information about social actions, social interactions, social emotions, social attitudes and social relationships". Social signals are non-verbal cues, such as face features, body gestures, and vocal behaviour. Other researchers have shown that social signals can be successful at predicting the behavioural outcomes of social situations such as speed dating. Hence, since personality affects behaviour, social signals should be correlated with personality. There are proposed methods to identify the big five personality traits for personality recognition, especially acoustic cues from speech. However, such methods are focused mainly on personality recognition by strangers (zero-acquaintance) and not accurate personality recognition. The research shows the difference between personality perception and personality recognition and demonstrate how to recognise personality accurately. Further research into personality recognition has unveiled a huge gap in personality research between computer science community and psychology community. Available corpora are built on stranger agreement (personality perception), and not on accurate personality judgement (personality recognition). Therefore, new corpus was collected based on accurate personality judgement model and experiments with the new corpus show that there is a correlation between the big five and social signals (acoustic cues). In addition, the research shows that social signals (acoustic cues) can be used to recognize all big five personality traits as opposed to perceiving them. The research has found that by providing valid and accurate personality data then social signals can be captured and it is possible to have accurate automatic personality recognition. The results demonstrate how social signals are identified, captured and analysed to recognize personality traits accurately from speech alone. This research anticipates its results to be of value to many HCI research areas such as healthcare and e-learning. Furthermore, personality recognition is a major issue in human-robot interaction (HRI), and this research will be of relevance to their future developments.
Degree
thesis:*- Name dc:type.qualificationname
- Ph.D
- Level dc:type.qualificationlevel
- doctoral
- Grantor dc:publisher.institution
- University of Sheffield
- Year dc:date.issued
- 2021
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- AlHammadi, Dina
- Advisor dc:contributor.advisor
-
- Moore, Roger
Rights
- Language dc:language.dcterms_iso639_2
- eng
Identifiers
dc:identifier.*- Identifier
- uk.bl.ethos.839236