Architecting dietary intake monitoring as a service combining NLP and IoT
MetadataShow full item record
AuthorBenítez-Guijarro, Antonio; Callejas Carrión, Zoraida; Noguera García, Manuel; Benghazi, Kawtar
natural language processinginternet of thingsnutrition e-coach
Benítez-Guijarro, A., Callejas, Z., Noguera, M. et al. Architecting dietary intake monitoring as a service combining NLP and IoT. J Ambient Intell Human Comput (2019). https://doi.org/10.1007/s12652-019-01553-2
SponsorshipDEP2015-70980-R of the Spanish Ministry of Economy and Competitiveness (MINECO) and European Regional Development Fund (ERDF); European Union’s Horizon 2020 research and innovation programme under grant agreement No 823907; COST Action IC1303 AAPELE
Currently there exist many tools that support monitoring and encouragement of healthy nutrition habits in the context of wellness promotion. In this domain, interfaces based on natural language provide more flexibility for nutritional self-reporting than traditional form-based applications, allowing the users to provide richer and spontaneous descriptions. Nonetheless, in certain circumstances, natural language records may miss some important aspects, such as the quantity of food eaten, which results in incomplete recordings. In the Internet-of-Things (IoT) paradigm, smart home appliances can support and complement the recording process so as to make it more accurate. However, in order to build systems that support the semantic analysis of nutritional self-reports, it is necessary to integrate multiple inter-related components, possibly within complex e-health platforms. For this reason, these components should be designed and encapsulated avoiding monolithic approaches that derive in rigidity and dependency of particular technologies. Currently, there are no models or architectures that serve as a reference for developers towards this objective. In this paper, we present a service-based architecture that helps to contrast and complement the descriptions of food intakes by means of connected smart home devices, coordinating all the stages during the process of recognizing food records provided in natural language. Additionally, we aim to identify and design the essential services that are required to automate the recording and subsequent processing of natural language descriptions of nutritional intakes in association with smart home devices. The functionalities provided by each of these services are ready to work in isolation, just out of the box, or in downstream pipeline processes, bypassing the inconveniences of monolithic architectures.