RESUMEN
With the rapid advancement of robotics technology, an increasing number of researchers are exploring the use of natural language as a communication channel between humans and robots. In scenarios where language conditioned manipulation grounding, prevailing methods rely heavily on supervised multimodal deep learning. In this paradigm, robots assimilate knowledge from both language instructions and visual input. However, these approaches lack external knowledge for comprehending natural language instructions and are hindered by the substantial demand for a large amount of paired data, where vision and language are usually linked through manual annotation for the creation of realistic datasets. To address the above problems, we propose the knowledge enhanced bottom-up affordance grounding network (KBAG-Net), which enhances natural language understanding through external knowledge, improving accuracy in object grasping affordance segmentation. In addition, we introduce a semi-automatic data generation method aimed at facilitating the quick establishment of the language following manipulation grounding dataset. The experimental results on two standard dataset demonstrate that our method outperforms existing methods with the external knowledge. Specifically, our method outperforms the two-stage method by 12.98% and 1.22% of mIoU on the two dataset, respectively. For broader community engagement, we will make the semi-automatic data construction method publicly available at https://github.com/wmqu/Automated-Dataset-Construction4LGM.
RESUMEN
Emotions influence our perceptions and decisions and are often felt more strongly in situations related to healthcare. Therefore, it is important to understand how both providers and patients express their emotions in face-to-face scenarios. An ontology is a way to represent domain concepts and the relationships between them in a polyarchical manner. We have created an ontological model called the Visualized Emotion Ontology (VEO) that expresses the semantic definitions and visualizations of 25 emotions based on published research. With VEO, we can augment patient-facing software tools, like embodied conversational agents, to improve patient-provider interaction in clinical environments.
RESUMEN
A análise do histórico de revistas científicas é um dos objetos principais de pesquisas nas áreas de bibliometria, cienciometria, informetria e webometria. Geralmente, essas análises procuram identificar o perfil cronológico dos artigos, autores e contextos editoriais das publicações. Quando entendida como uma organização inserida em um sistema de ciência, tecnologia e inovação (CT & I), uma revista científica é um agente de memória organizacional, que dissemina e promove conhecimento. Assim, além de análises sobre seu contexto editorial, é relevante verificar outros fatores que posicionam a revista no sistema de CT & I ao qual se refere. Uma das formas de tratar essa questão se dá pela combinação de métricas da informetria com análises oriundas da engenharia e da gestão do conhecimento. Neste artigo, aplica-se um modelo multidisciplinar com essa natureza, para verificar a base de conhecimentos criada pela Revista Engenharia Sanitária e Ambiental (Revista ESA) no sistema brasileiro de CT & I. Foram analisados os 333 artigos publicados entre agosto de 2004 e dezembro de 2012, o perfil curricular dos 816 autores, bem como o perfil de financiamento em CT & I realizados nos fundos setoriais em temáticas afins à revista, no mesmo período das publicações. Os resultados dessas análises foram verificados por especialistas no domínio das ciências ambientais e engenharia sanitária, com experiência no histórico da revista. Os resultados indicam que o perfil de conhecimentos produzidos pela Revista ESA guarda correspondência com os critérios de financiamento federais para CT & I, evidenciando o papel que a revista representa como formadora de uma base de conhecimento científico em engenharia sanitária e ambiental.
One of the main goals of bibliometrics, scientometrics, informetrics and webometrics is to analyze the history of scientific journals. Usually these studies analyze journals history, authors' profiles and how publications have evolved over time. In a broader view, a scientific journal can be thought of as a memory agent that promotes and disseminates knowledge in a science, technology and innovation system (STI). This brings other possibilities of understanding the role of scientific journals in STI systems. We address this challenge by combining informetrics with knowledge engineering and management techniques. In this article, we apply a multidisciplinary model to verify the knowledge base created by the Brazilian Scientific Journal called "Engenharia Sanitária e Ambiental" (ESA) in the Brazilian STI system. We analyzed the 333 articles published between August 2004 and December 2012. We also studied the national database of curricula in Brazil to analyze the profiles of the 816 authors and a national database, to check for public funding in subjects published in ESA. We concluded that both the knowledge published in ESA and the areas funded by national grant in Brazil have evolved in a similar way. This indicates that ESA plays a significant role as a memory agent in environmental and sanitary engineering in Brazil.
RESUMEN
BACKGROUND: The development of decision support systems for nursing has been limited by difficulties in defining and representing the nursing knowledge base and by a lack of knowledge about how nurses make decisions. However, the current trends suggest that many of the formidable technological and conceptual challenges associated with representing nursing knowledge can be overcome, and that decision support systems developed using knowledge engineering can be used to significantly improve nursing practice. OBJECTIVE: The aim of this study was to develop a diagnostic nursing decision support system for inpatients with type 2 diabetes mellitus, and to elucidate the advantages and disadvantages of applying a knowledge engineering approach to its development. METHODS: To acquire the relevant knowledge, a literature review was used to iteratively establish nursing diagnoses and clinical assessment criteria. Twenty-five NANDA diagnoses and 145 clinical assessment variables were structured into a criteria table used as the knowledge base for a prototype system. To investigate the responses of nurses to knowledge presentation, 27 nurses from inpatient and ambulatory settings and the graduate course of nursing informatics were recruited and a scenario-based preliminary evaluation was performed. RESULTS: A prototype for inpatients with type 2 diabetes mellitus was developed as a Web-based stand-alone system. It automatically generated suggested nursing problem lists based on the input data, and also provided detailed explanatory information for each item on the list. We describe the advantages and disadvantages of the development approach used and discuss the users' impressions of the system.