Download presentation
Presentation is loading. Please wait.
Published byGervase Sharp Modified over 7 years ago
1
PaaSport Offerings Recommendation Layer & Semantic Models
Nick Bassiliades International Hellenic University (IHU) Final Review Meeting - 12/01/ Brussels
2
PaaSport Architecture
Final Review Meeting - 12/01/ Brussels
3
Semantic Models & PaaS Offering Recommendation Layer
PaaSport Semantic Models (IHU) conceptual and modelling pillars of the marketplace infrastructure. Semantic annotation of the registered PaaS offerings and the deployed applications profiles. PaaS Offering Recommendation Layer (NUIG, IHU) implements core functionalities offered by the PaaSport Marketplace Infrastructure. PaaS offering discovery, recommendation and rating. Final Review Meeting - 12/01/ Brussels
4
The role of the PaaSport Semantic Models
Common vocabulary for the various modules of the system PaaSport Offering Recommendation layer Matchmaking is performed using concepts and properties of the offering and application ontologies. Persistence layer The DB schema follows exactly the conceptual model of the ontologies; No mismatch. PaaS Offering profiles stored in the DB are mapped onto the Semantic Model layer (RDF graph) using a relational-to-ontology mapping tool (D2RQ platform). Adaptive front-ends layer The UIs for managing application and PaaS offering semantic profiles use concepts and properties from the semantic models. Final Review Meeting - 12/01/ Brussels
5
Semantic Models Objective
Define a formal pattern for modelling various aspects of PaaS concepts (offerings, applications and SLAs) Basic modelling requirements User roles, e.g. providers, clients Conceptual modelling of properties/attributes for PaaS Further descriptions of properties through e.g. n-ary relations, container classes, etc. Units of measurements Easy to reuse/inject any existing domain vocabulary Matchmaking-oriented attributes, e.g. value ranges Easily extendable Well-defined steps to be followed Final Review Meeting - 12/01/ Brussels
6
PaaSport Semantic Models
Extension of the DUL upper ontology (DOLCE+DnS Ultralite) Well established and widely used Semantic Interoperability with other similar research efforts (e.g. SSN) Extensibility Adding new concepts and parameters related to PaaSport offerings and applications Without compromising the Table Schema of the Persistence Layer. Final Review Meeting - 12/01/ Brussels <Event/Title> <date> <Location>
7
PaaSport Semantic Models Overview
Final Review Meeting - 12/01/ Brussels
8
A PaaS Offering Example
Final Review Meeting - 12/01/ Brussels
9
An Application Request Example
Final Review Meeting - 12/01/ Brussels
10
PaaS Concepts & Parameters
Final Review Meeting - 12/01/ Brussels
11
PaaS Offerings Recommendation layer Objective
To allow the user to select the most appropriate PaaS offering published at the PaaSport marketplace based on the semantic description and deployment profile of the business applications that the user wishes to deploy in the Cloud. Final Review Meeting - 12/01/ Brussels
12
Final Review Meeting - 12/01/2017 - Brussels
PaaS Offerings Recommendation layer Architecture – Component Interaction Final Review Meeting - 12/01/ Brussels
13
Matchmaking and Ranking algorithm
The heart of the PaaS Offering Recommendation Layer The algorithm selects, scores and ranks the most appropriate PaaS offerings that best match the requirements of the application a developer wants to deploy. Requirements / Objectives Efficiency / scalability Extensibility When the semantic model is extended with new offering / applications instances, concepts and parameters. The algorithm should not (or minimally) change. Final Review Meeting - 12/01/ Brussels
14
Matchmaking and Ranking algorithm
GUI Initialization GetOfferings Input Check Functional For each concept, offering and parameter Run SPARQL query Keep offering or not Rank Non-Functional Retrieve parameter values Score parameter values Sort Results Application Requirement Instance Offerings Concepts Functional Parameters Non-Functional Parameters Coefficients Difference Functions Functional? Steep-ness Factor PaaS Offering profiles Offerings Offerings + Score Final Review Meeting - 12/01/ Brussels
15
Final Review Meeting - 12/01/2017 - Brussels
Scoring PaaS Offerings How much do we care for better parameter values? OfferScore= 𝑝∈𝑁𝑜𝑛𝐹𝑢𝑛𝑐𝑡𝑖𝑜𝑛𝑎𝑙 𝑐𝑜𝑒𝑓𝑓 𝑝 ∗ 𝑃𝑎𝑟𝑆𝑐𝑜𝑟𝑒 𝑝 Final Review Meeting - 12/01/ Brussels
16
Final Review Meeting - 12/01/2017 - Brussels
Implementation Java for programming all components Including the matchmaking algorithm Jena Framework for handling Ontologies / RDF Querying via SPARQL query language for RDF D2RQ for mapping relational DB to RDF triples Mapping rules in the D2RQ Mapping Language Spring framework for retrieving the application requirement from the persistence layer. Final Review Meeting - 12/01/ Brussels
17
Final Review Meeting - 12/01/2017 - Brussels
Matchmaking Workflow When new offerings are added, the mapping procedure is incrementally run to update the KB of PaaS offerings. Final Review Meeting - 12/01/ Brussels
18
Algorithm Performance
Query time scales linearly to the number of: Offerings Requirement parameters Final Review Meeting - 12/01/ Brussels
19
The PaaSport Rating and Selection Components
Rating Component The user can rate PaaS Offerings based on his criteria. Selection Component The user can sort the PaaS Offerings based on their ratings and his priorities/criteria. Criteria Quality of Service Quality of Support Maintenance Activity SLA Reliability Value of Money Final Review Meeting - 12/01/ Brussels
20
…so How Does Rating & Selection Work?
Final Review Meeting - 12/01/ Brussels
21
Final Review Meeting - 12/01/2017 - Brussels
Technical Example User priorities {"QualityOfService": 4, "QualityOfSupport": 5, "MaintenanceActivity": 2, "SLAReliability": 1, "ValueForMoney": 3}, Service 1 {"QualityOfService": 2, "QualityOfSupport": 3, "MaintenanceActivity": 2, "SLAReliability": 1, "ValueForMoney": 3}, Service 2 {"QualityOfService": 3, "QualityOfSupport": 2, "MaintenanceActivity": 5, "SLAReliability": 1, "ValueForMoney": 1}, Service 3 {"QualityOfService": 2, "QualityOfSupport": 3, "MaintenanceActivity": 2, "SLAReliability": 1, "ValueForMoney": 3} Service 4 {"QualityOfService": 4, "QualityOfSupport": 5, "MaintenanceActivity": 4, "SLAReliability": 2, "ValueForMoney": 2} Output ranking scores attached to the services. [1] "service 1 score " [2] "service 2 score " [3] "service 3 score " [4] "service 4 score " Final Review Meeting - 12/01/ Brussels
22
Final Review Meeting - 12/01/2017 - Brussels
Relevant Publications F. Ahmadi Zeleti, I. Hassan, S. Abbas, A. Ojo and L. Porwol, "Architecting the Recommendation Layer of a Platform-as-a-Service e-Marketplace", Proc. 11th International Joint Conference on Software Technologies - Volume 1: ICSOFT-EA, pp DOI: / N. Bassiliades, M. Symeonidis, G. Meditskos, E. Kontopoulos, P. Gouvas and I. Vlahavas, "A semantic recommendation algorithm for the PaaSport platform-as-a-service marketplace", Expert Systems with Applications (IF: 2.981), Elsevier, Vol. 67, Jan. 2017, pp , N. Bassiliades, M. Symeonidis, G. Meditskos, E. Kontopoulos, P. Gouvas and I. Vlahavas, “PaaSport Semantic Model: An Ontology for a Platform-as-a-Service Semantically Interoperable Marketplace”, under review. Contact: Nick Bassiliades International Hellenic University & Aristotle University of Thessaloniki Final Review Meeting - 12/01/ Brussels
Similar presentations
© 2025 SlidePlayer.com Inc.
All rights reserved.