Publicação em atas de evento científico
Data sense platform
Mariana Dias (Mariana Dias); Joao C Ferreira or Joao Ferreira (Ferreira, J.); Rui Maia (Maia, R.); Rui Maia (Rui Maia); Rui Maia (Maia, R.); Rui Maia (Rui Maia); Pedro Santos (Santos, P.); Pedro Santos (Pedro Santos); Ricardo Ribeiro (Ribeiro, R.); Maria Major (Major, M.); et al.
Proceedings of IASTEM International Conference
Ano
2019
Língua
Inglês
País
Brunei
Mais Informação
Abstract/Resumo
The current manual or semi-automatic document preservation process suffers from various problems that particularly affect the handling of confidential or sensitive information, such as the identification of sensitive data in documents requiring human intervention that is costly and propense to generate error, and the identification of sensitive data in large-scale documents does not allow an approach that depends on human expertise for their identification and relationship. DataSense will be highly exportable software that will enable organizations to identify and understand the sensitive data in their possession in unstructured textual information (digital documents) in order to comply with legal, compliance and security purposes, identify and classify and relate sensitive data (Personal Data) present in large-scale nonstructured information in a way that allows entities and/or organizations to understand it without calling into question security or confidentiality issues, and allowing companies that focus on their clients to better understand their profile from information collected from sensitive data or consent data or algorithms. The Data Sense project will be based on 3 key layers using the current potential of NLP technologies and the advances in machine learning (NER), Disambiguation and Coreferencing (ARE) and Automatic Learning and Human Feedback. It will also be characterized by the ability to learn from human feedback automatically, correcting and iteratively improving the AI model that supports it.
Agradecimentos/Acknowledgements
--
Palavras-chave
s- Sensitive Data; Natural Language Processing; Text Mining; Named Entities Recognition; Co-reference Resolution.;
  • Ciências da Computação e da Informação - Ciências Naturais
Registos de financiamentos
Referência de financiamento Entidade Financiadora
UID/MULTI/0446/2013 Fundação para a Ciência e a Tecnologia
UID/GES/00315/2013 Fundação para a Ciência e a Tecnologia