| Yhteenveto: | The coming of the Big Data era has posed great challenges to the traditional de- cision support systems, which are unable to effectively leverage unstructured data, necessi- tating more flexible and adaptable approaches. Originating from the same acknowledgment expressed in the Value from Public Health Data with Cognitive Computing project, this study introduces a text-based approach to designing decision support systems and evaluates its practicality, utility as well as its advantages in facing these challenges. The potential ben- efits from leveraging Semantic Web technologies as a driving force and in improving the performance of such systems were also investigated. For assessing the validity of the ap- proach in practice, two proof-of-concept prototypes were developed in succession.
Theoretical analysis showed that a text-based decision support system is fully capable of alleviating the difficulties faced by traditional systems in utilizing unstructured textual data in the decision-making process. On the other hand, the implementations of the prototypes demonstrated the possibility of employing large-scale and well-structured ontologies like SNOMED-CT as the basis for knowledge representation, resulting in performance gain. At the same time, the application of the proposed semantic relevance measure was shown to further enhance the derivation of relevant information. While additional and more conclusive evaluations are needed, the study proved that a text-based ontology-driven decision support system is feasible and worthy of further research.
|
|---|