Events at public beaches are one of the most popular recreational activities of local communities and international visitors in all places around the world. Amongst others, the beach safety management in protected areas needs support for continuous analysis and decision making on incidents at the beach areas. There is a lack of available standard models to assist data scientists to represent analytics models including different spheres of interplay domain variables related to beach safety management. Using the Design Science Research Methodology (DSRM), we developed ontological models that facilitate a unified representation and maintenance analytics models. We contribute to the ontology design theory for analytics models underlying analytics systems for the beach safety management domain. Our research findings can be used in the general class of ontology design problem for analytics systems in practice.
Double Revolving field theory-how the rotor develops torque
Role of ontologies in beach safety management analytics systems
1. Role of Ontologies in Beach Safety
Management Analytics Systems
(Paper order: 1783)
1
2. • Outline
• Research context
• Research problem and objective
• Research methodology
• Lessons learned (key takeaways)
• F&Q
2
3. • Public beaches are one of the most popular recreational activities of
local communities
• a vibrant public space for locals and international
• Life-threatening injuries
• International Life Saving Federation (ILSF) reports that 1.2 million
people around the world annually lose their lives due to drowning in
open water such as beaches, sea, lakes, and rivers.
• unfamiliarity with surfing conditions
• poor swimming skills
• weather condition
• disorientation in coastal areas
3
Research context
4. • Smart Beaches (IoT & Analytics
enabled)
• Beach safety management agencies leverage a
wide range of information technologies such as IoT
(Internet of Things) devices, drones, wearable
sensors, outdoor cameras, and mobile applications
• Continuous monitoring of beach space to
detect hazards and alarm risks in a real-time
fashion
• Massive data (aka. Big Data) collected from these
technologies
4
Research context
5. Analytics models(Watson 2014)
• Descriptive analytics models
• Sum of drowning
• Average of shark attacks per month
• Average of beachgoers
• Example: bar charts, pies
• Predictive analytics
• Project future possibilities to answer questions
• what type of incidence is likely to happen at the beach in a
public holiday?
• Example: regression and machine learning
• Prescriptive analytics
• Decision making
• What actions should be taken to avoid drowning incidence
during summer break?
• Example: simulation and decision models
5
Analytics models, i.e., descriptive, predictive, prescriptive
Beach safety management domain
Research context
6. • Knowledge gaps in developing Analytics Information Systems (in smart beach domain)
•Communicative and knowledge intensive exercise
•What are incorporating variables informing analytics models, especially beach safety management domain?
•Efficient and consistent knowledge flow about analytics models among data scientists
•An overarching view that can pull together the various domain variables describing analytics models
•Also called “Data and model disparity problem” in analytics
Analytics models
(descriptive, predictive, prescriptive)
Analytics systems
Beach safety management domain
Data science team
Research problem and objective
7. • A potential solution: Ontologies (e.g., domain conceptualisation)
• A systematic explanation of being (Kishore 2004)
• Interoperability, and bridging knowledge among stakeholders
• Structure and codify knowledge about concepts, relationships, and axioms/constraints in a specific context
• Computational format, competency questions (CQ)
• CQ, what risk factors occur at beach?
• is represented by: What [V1] [OPE] [V2]?
• V1, i.e., risk factors and V2, i.e., beach, are variable expressions
• OPE, i.e., occur, is an object property expression
• Research objective
• To develop an ontology for beach safety management domain to inform data scientists of operational variables and
data to incorporate into analytics models.
7
Analytics models
(descriptive, predictive, prescriptive)
Analytics systems
Beach safety management domain
Data science team
Research problem and objective
8. • Research methodology
• Design science research methodology-DRSM (Gregor and Hevner 2013)
• Smart beach ontology artifact
• Design cycles and iterative artefact refinements
8
Research methodology
9. • Sample Smart Beach Ontology (SBO) artifacts
• Ontologies that capture knowledge about analytics models for the beach safety management
domain
• Domain variables and relationships
9
Research methodology – design cycles
10. 10
Instead of struggling to understand analytics mode in ad-hoc way,
Smart Beach Ontology (SBO) artifacts act as a guidance or
conceptual model to inform data science team about what variables
are and how they might be related..
Research methodology – design cycles
Data science team
14. 14
Smart Beach Ontology as a point of reference for transforming, either manual or automatic, analytics
queries/requirements to analytics information systems
Research methodology – design cycles
15. • Key takeaways in designing ontologies
for analytics systems
• Trade-off among ontology design principle
• Design principles: completeness, correctness, expandability,
conciseness, consistency, clarity
• For example: generic and less-detailed ontology vs. a comprehensive
one versatile
• Reciprocal benefit
• Whilst SBO narrows its focus on the standardization and integration of
analytics models at beach safety agencies, analytics models can help to
verify knowledge captured by the ontology
• Noisy variables might be removed from the ontology after some
analytics model development
• Variable drift
• Rapidly changing environments of new type of incidents, new rescues
actions, and new visitor hazardous behaviors at beach space inevitably
results in a change of meaning for variables
• The notion of Rescue can be completely changed from human lifesaver
to fully automatic drone lifesaver
15
Lessons learned (key takeaways)
16. 16
Mahdi Fahmideh, PhD in Information Systems
University of Southern Queensland, Australia,
E: Mahdi.Fahmideh@usq.edu.au , M: +61406052400
Notes de l'éditeur
Photo source: smartbeaches.com.au, with kind permission