36 37 35 38 39 estamo B Relleno Desmonte 5000m3 8000m3 25% roca 75% tierra Complejidad: Sencilla CASO DE PRUEBA BComplejidad: Elevada Madeira Lisboa Data Mining Cloud Computing Artificial Intelligence 8th International Conference on Cloud Computing and Services Science, CLOSER 2018 March, 2018 — Funchal (Madeira, Portugal) A Proposal for the Specification of Data Mining Services in Cloud Computing Manuel Parra-Royón J.M. Benítez-Sánchez ! Distributed Computational Intelligence and Time Series Lab Soft Computing and Intelligent Information Systems Department of Computer Science and Artificial Intelligence University of Granada (Spain) 

8th International Conference on Cloud Computing and Services Science, CLOSER 2018 March, 2018 — Funchal (Madeira, Portugal) Outline! ! Introduction Cloud computing: Services in cloud Services definition languages for cloud computing Services definition languages for data mining and cloud computing Cloud computing features + DataMining / Machine Learning Service definition for Cloud Computing Data Mining: dmcc-schema Deploy of dmcc-schema on a real scenario Proof of concept Conclusions 

8th International Conference on Cloud Computing and Services Science, CLOSER 2018 March, 2018 — Funchal (Madeira, Portugal) Manuel Parra! PhD Student & Researcher BigData Time Series Data Mining University of Granada 
 Soft Computing and Intelligent Information Systems Distributed Computational Intelligence and Time Series DiCITSlab Distributed Computational Intelligence and Time Series Lab 

8th International Conference on Cloud Computing and Services Science, CLOSER 2018 March, 2018 — Funchal (Madeira, Portugal) Rise of cloud computing on services of Machine Learning (MLaaS): Amazon AWS, Microsoft Azure,Google Cloud Platform, IBM Watson, … ! Providers offer its own catalogue of services with their particular specification details. ! A cloud computing data mining service:
 Not just run experiments or algorithms Must consider: authentication, catalog, pricing, etc. Key aspects in cloud computing Introduction 

8th International Conference on Cloud Computing and Services Science, CLOSER 2018 March, 2018 — Funchal (Madeira, Portugal) Storage as a Service (DropBox, GoogleDrive, …) Plataform as a Service (Heroku, OpenShift,…) Data Bases as a Service (Oracle, MySQL, …) Infraestructure and computing resources as a Service (OpenStack, OpenNebula, Eucaliptus, …) Data as a Service (Spark, Hadoop, etc.) … Cloud computing: Services in cloud UTILITY model — Pay for the service 

8th International Conference on Cloud Computing and Services Science, CLOSER 2018 March, 2018 — Funchal (Madeira, Portugal) Cloud computing: Services in cloud NIST defines 5 properties for services on cloud computing ! ‣Easy to access and to consume ‣Easy elasticity ‣Shared resources ‣Auto-service discoverable ‣Measurable - Quantifiable 

8th International Conference on Cloud Computing and Services Science, CLOSER 2018 March, 2018 — Funchal (Madeira, Portugal) How to define or describe a cloud service ? Cloud computing: Services in cloud Business EndPoints Discovery Costs Authentication Constraints Parameters Resources … Entities 

8th International Conference on Cloud Computing and Services Science, CLOSER 2018 March, 2018 — Funchal (Madeira, Portugal) Cloud computing: Services in cloud 

8th International Conference on Cloud Computing and Services Science, CLOSER 2018 March, 2018 — Funchal (Madeira, Portugal) Semantic Syntactic OWL-S WSMO SA-WSDL … WSDL UDDI SoaML OpenML … Not fully addressed with Service Oriented Architecture (SOA) Services definition languages for cloud computing 

8th International Conference on Cloud Computing and Services Science, CLOSER 2018 March, 2018 — Funchal (Madeira, Portugal) MachineLearning-schema (mls) MEX (mex) … Specific OWL LinkedUSDL … General Without support Cloud Computing Virtually support for Cloud Computing services Complexity Service definition languages for data mining and cloud computing 

8th International Conference on Cloud Computing and Services Science, CLOSER 2018 March, 2018 — Funchal (Madeira, Portugal) USDL, LinkedUSLD, SA-WSDL, OWL, UDDI…. Focused in Data Mining Focused in general services definition Lack of Cloud Computing features Data Mining + Cloud Computing + Service definition Gap Service definition languages for data mining and cloud computing 

8th International Conference on Cloud Computing and Services Science, CLOSER 2018 March, 2018 — Funchal (Madeira, Portugal) Machine-Learning as a Service (ML-as-a-Service) ! Service provider includes: Compute costs (hours of processing) Computing resources (CPU, RAM, …) Storage cost (data and models) Algorithms, parameters, … … Cloud computing features + DataMining / Machine Learning 

8th International Conference on Cloud Computing and Services Science, CLOSER 2018 March, 2018 — Funchal (Madeira, Portugal) Our proposal: dmcc-schema It works between a generic cloud service and data mining services. ! It proposes a schema and vocabulary by using LinkedData to unite both facets (cloud computing and data mining) Service definition for Cloud Computing Data Mining 

8th International Conference on Cloud Computing and Services Science, CLOSER 2018 March, 2018 — Funchal (Madeira, Portugal) Full data mining algorithms experimentation and execution Define multiple APIs and entry points of interaction Price specification of the services details Service catalog discovery Authentication methods Roles of interaction Software license agreement Our proposal dmcc-schema + LinkedData Service definition for Cloud Computing Data Mining 

8th International Conference on Cloud Computing and Services Science, CLOSER 2018 March, 2018 — Funchal (Madeira, Portugal) What about Linked Data? ! Linked Data is based on RDF, represents data as a graph:
 a collection of resources with relationships to one another Linked Data is Sharable, extensible, and easily re-usable. Linked Data describes the meaning of data ("semantics") separately from specific data structures ("syntax" or "formats") Service definition for Cloud Computing Data Mining 

8th International Conference on Cloud Computing and Services Science, CLOSER 2018 March, 2018 — Funchal (Madeira, Portugal) dmcc-schema + LinkedData mls: Machine Learning Schema waa: Web API Autenticación skos: Simple Knowledge Organization System Namespace sla: LinkedDataUSDL module SAL gr: Good Relations foaf: Friend of a Friend … Linked Open Vocabularies (LOV) Service definition for Cloud Computing Data Mining 

8th International Conference on Cloud Computing and Services Science, CLOSER 2018 March, 2018 — Funchal (Madeira, Portugal) http://dicits.ugr.es/occml/ Create a new service K-means with dmcc-schema Deployed with Turtle or JSONLD Service definition Service composition Deploy of dmcc-schema on a real scenario Turtle 

8th International Conference on Cloud Computing and Services Science, CLOSER 2018 March, 2018 — Funchal (Madeira, Portugal) Create a new service K-means with dmcc-schema Defined with Turtle or JSONLD Parameters IN/OUT Parameters IN Data mining model Deploy of dmcc-schema on a real scenario Turtle 

8th International Conference on Cloud Computing and Services Science, CLOSER 2018 March, 2018 — Funchal (Madeira, Portugal) Develop of OCCML data mining serveless platform http://dicits.ugr.es/occml/ MachineLearning-as-a-Service in your own infrastructure. Proof of concept 

8th International Conference on Cloud Computing and Services Science, CLOSER 2018 March, 2018 — Funchal (Madeira, Portugal) OCCML: Service integration and discovery OCCML: Service composition Data Mining / Artificial Intelligence Service definicion dmcc-schema (Turtle RDF / JSONLD) Data Mining Algorithms Service Catalog SPARQL API OCCML: Executor OCCML Core occml.balancer occml.resourcescale occml.mlinterface occml.storage occml.ServiceProcessor occml.eventor Interoperated from: RESTFul API, Integrated in your code, … Proof of concept 

8th International Conference on Cloud Computing and Services Science, CLOSER 2018 March, 2018 — Funchal (Madeira, Portugal) Conclusions Linked Data dmcc-schema solves gap between data mining and cloud computing services definition Validation of dmcc-schema is probed on OCCML platform working with more than 50 algorithms as a service (included data-processing) With dmcc-schema, services published can be discovered and consumed easily by a programatic agent, users, or cloud computing broker 

Thanks ! Questions ? Manuel Parra-Royón Twitter: @manugrapevine Email: [email protected] ! Distributed Computational Intelligence and Time Series Lab Soft Computing and Intelligent Information Systems Department of Computer Science and Artificial Intelligence University of Granada (Spain) DiCITSlab Distributed Computational Intelligence and Time Series Lab