STAG4Covid-19 is able to semantically categorize texts according to a given categories taxonomy and to extract from the very text the relevant keywords including Named Entities (NE), i.e. health-related entities. We will use as categories the UMLS Concepts Unique Identifiers (CUIs). Standard NE Recognition techniques will be extended with domain-specific technologies (e.g. MetaMap, cTAKES, CLAMP, RapTAT). STAG4Covid-19 will provide the mean to semantically couple NEs to their contexts, which are in turn classified according to CUIs. This will straightforwardly lead to the normalization of NE.
n medical research papers or in user-generated content (UGC), health-related entities (e.g. findings, symptoms, diseases, diagnoses, and medications) are mentioned very often in a not normalized and not semantically interoperable form. This is because these entities do not relate to concepts represented in a controlled vocabulary. This leads, today, to reduced efficiency in the circulation of the information among the scientific community and, in a near future, it will prevent the exploitation of the UGC in tasks such as the pharmacovigilance, that will be of pivotal relevance when a partially tested vaccine or a new drug will be released to the bulk of the population.
Adaptation of STAG scalable architecture. Dissemination of the solution in the scientific community.
The clinicians, the hospitals, the scientific community in general and the public health monitoring institutions are the target of STAG4Covid-19. For example, the concept of heart attack will be made automatically recognizable among different definitions and list of symptoms. Access to the services will be granted to single users. There is no intrinsic limitation to the number of users to be served.
STAG is a fully parallelizable technology. The number of Front End servers that will be exposed to the users, as well as the number of the Back End servers that will process their requests, can dynamically scale with the volume of the requests.