Research

The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man.

George Bernard Shaw

My driving belief is that technology as a whole should be used in the service of humanity. Applied properly it will enhance health, education, safety, justice, and economic equality. Many of these emerging social technologies will rely heavily on socially enriched human-machine interaction – sufficiently rich that it allows for analysis, understanding and communication. In particular, analysis focusing on socio-cultural dialogue framework, among others, would greatly aid in the future advances of socially aware conversational agents especially in increasing cross-culture acceptance.

Research Interests

My research interests span in the areas of Human-Machine Interaction, specifically Affective Computing. A common thread in my research is in understanding the theory and design of scalable and independent architecture of socio-emotional models in conversational agents. A system that underlies a purely affective model may still be socially incompetent.  A system infused with social awareness ideally would integrate a culture model. Therefore, in a smaller scope, I am interested in researching and developing culture-sensitive emotive interfaces, which deals with interdisciplinary fields such as cross-cultural pragmatics, politeness strategies, empathy and conversational grounding, an upcoming field which is still in its infancy. This is motivated by the fact that the perception and expression of emotions vary among people from different cultures, and therefore the designs of emotive computer interfaces should be tailored to conform to the different needs according to cultural background. Being born and raised in a multicultural environment, and later  stepping into an intercultural marriage, in itself I believe, will be a great benefit for this research.

Background and Current Work

Despite the strong evidence for cross-cultural consistency in emotion appraisal processes and perceptions, cultural differences still exist. Even within a particular culture, there are cultural dialects. Mainstream affective applications that are developed with a general framework of emotion entrenched in the Western hegemony (individualistic culture) may not be suitable to be adopted in an environment belonging to collectivist culture (East). Differences in culture could cause negative judgments that bring conflicts and misunderstandings, which hinder the process of successful interaction. A simple example would be accepting and rejecting compliments; while Americans and Europeans respond to compliments by accepting them, Asians (not just limited to Chinese) frequently responded with rejection followed by self-denigration – it is seen as humility, which greatly molded the East Asian value system. Thus, cross-cultural varieties in spoken dialogue (and other modalities) are vital in order to move beyond stereotypes and can strongly aid in the future advances of socially aware conversational agents. Theoretical framework positing dimensions on cultural varieties would serve as guidelines to model the varieties across cultures.

Currently I am involved in modeling emotion for a task-independent robotic agent by integrating a module of needs. Very few studies on computational model of emotions  offer  task-independence. Even fewer have linked needs and emotions. Why needs, one may ask. It is a well established fact that emotions drive motivations. Likewise an affect system is influenced by a motivation (need-based) system. Need deficiencies or fulfillment are experimental emotions – the lower the need satisfaction, the more the related negative emotion is amplified, and vice versa. To demonstrate this, we propose an emotion framework for a task independent autonomous agent by integrating a module of needs. The need framework, called NEMO (Need-inspired Emotional System) is based on Abraham Maslow’s motivation theory (hierarchy of needs) and appraisal theory. Nemo was tested as the underlying emotion model for an autonomous robot that controls electrodomestic devices . A demo (in Spanish) is available. NEMO will be enhanced with a culture model and would be the underlying model for a HIFI conversational agent that was previously developed – as the next prototype.

Previous Projects

Emotional prosody modeling for a Malay TTS

Culture-sensitive TTS

Emotions identifications in voice

Leave a Reply

De Madrid..con amor..