Project Title

Funding Body and Scheme

Project No

Start Date – End Date

Total Budget

AgentDSM-Eval: Developing a Framework on Evaluating Domain-specific Modeling Languages for Multi-agent Systems

TUBITAK ARDEB 1001

FCT

115E591

01.03.2017 – 01.03.2020

326.980 TL (~$91,000)

Supporting the Interoperability between Domain-specific Agent Modeling Languages with Horizontal Model Transformations

Ege University

16-UBE-001

01.06.2016 – 01.06.2018

38.950 TL (~$13,200)

Development of an Agent based Alternative Path Selection Method for Peer-to-Peer Video Streaming Applications

TUBITAK ARDEB 1001

111E022

01.10.2011 – 01.10.2013

177.078 TL (~$99,000)

A Domain Specific Modeling Language for Semantic Web enabled Multi-agent Systems

TUBITAK ARDEB 1001

ARRS

109E125

01.03.2010 – 01.03.2013

129.144 TL (~$86,000)

Assessment and Performance Measurement of Semantic Service Discovery Methods on Mobile Platforms

Ege University

2009/UBE/003

01.11.2009 – 01.11.2012

14.000 TL (~$9,300)


AgentDSM-Eval: Developing a Framework on Evaluating Domain-specific Modeling Languages for Multi-agent Systems

The Scientific and Technological Research Council of Turkey (TUBITAK) Academic Research Funding Program Directorate (ARDEB) Electrical, Electronics and Informatics Research Grant Committee (EEEAG) and the Portuguese Foundation for Science and Technology (FCT) funded bilateral research project (TUBITAK ARDEB Grant no: 115E591) (acceptance rate of the related bilateral project call: 6%)

        

Software agents and multi-agent systems (MASs) are recognized as both useful abstractions and effective technologies for modeling and building of autonomous composite systems. However, autonomous, responsive and proactive nature of agents makes development of agent-based software systems more complex than many other systems. Furthermore, both internal agent behavior model and interactions within the agent organizations become even more complex and hard to implement when taking into account the varying requirements of different agent environments. It is obvious that working in a higher abstraction level is of critical importance for the development of MASs since it is almost impossible to observe code level details of MASs due to their internal complexity, distributedness and openness. In order to master the abovementioned problems of developing MASs, agent-oriented software engineering (AOSE) researchers define various agent metamodels which include fundamental entities and relations of agent models. By enriching those metamodels with some defined syntax and semantics, researchers also propose domain-specific modeling languages (DSML) for facilitating the development of MAS. Domain-specific languages (DSLs) have notations and constructs tailored toward a particular application domain (e.g. MAS). The end-users of these languages have knowledge of the problem domain, but usually they have little programming experience. DSMLs further raise the abstraction level, expressiveness and ease of use, since models are specified in a visual manner and they represent the main artifacts instead of software codes. Apparently, the use of a DSML helps to the model-driven development (MDD) of MASs. MDD aims to change the focus of software development from code to models, and hence many AOSE researchers believe that this paradigm shift introduced by MDD may also provide the desired abstraction level and simplify the development of complex MAS software. Originating mainly from pre-defined abstract syntaxes, various DSLs / DSMLs (DS(M)Ls) for agent systems have been developed recently. Many of these languages support modelling both static and dynamic aspects of agent software from different MAS viewpoints including agent internal behaviour model, interaction with other agents, use of other environment entities, etc. In addition, some of those languages were also enriched with appropriate integrated development environments (IDEs) in which both modelling and code generation for system-to-be–developed can be performed properly. To be effective, the proposed agent DSMLs need to meet the various stakeholder concerns and the related quality criteria for the corresponding MASs. Unfortunately, very often the evaluation of the DSML, especially covering the language components, generated artifacts, and the use of the DSML during design and implementation of agent-based systems, is completely missing or has been carried out with an idiosyncratic approach. If the DSMLs are not well-defined then implicitly this will have an impact on the quality of the MASs. Hence, in this bilateral project, our aim is to develop an evaluation framework which provides the systematic assessment of existing or newly defined DSMLs for MASs. This approach will also pave the way for making the systematic comparison and selection of MAS DSMLs possible. The bilateral project includes a collaborative work between Turkish and Portuguese research groups from Ege University and Universidade Nova de Lisboa respectively. Co-directors of the project are Assoc. Prof. Dr. Geylani Kardas from Ege Universiy and Asst. Prof. Dr. Vasco Amaral from Universidade Nova de Lisboa. Following the investigation of in-use DSML evaluation approaches, metrics related to both qualitative and quantitative assessment of agent DSMLs will be produced. Formalization of these metrics will be based on the adaptation of the currently available evaluation criteria and then generation of the new criteria which enables to the fair assessment of the MAS DSMLs by taking into consideration of various viewpoints of agent systems, e.g. agent internal architecture, MAS organization and interaction between agents. Derivation of both the evaluation approach and metrics in question will be originated from the broad investigation of the features presented by the various MAS DSMLs. Available IDEs for MAS DSMLs will also be experienced and evaluated in order to construct an extensive framework which is powerful enough to support the evaluation of agent DSMLs from different aspects. Evaluation of those DSMLs will be conducted via a multi case study research approach in which the generation of agent-based software systems for different domains is considered. It also covers the assessment of the tool artifacts, generation performance and user’s evaluation. In the process of the evaluation, all phases of MAS development are elaborated. In addition, the evaluators’ feedback is gained both by developing case studies and by working on questionnaires. The main outcome of the project will be a complete evaluation framework, called AgentDSM-Eval shortly, which provides the systematic assessment of MAS DSMLs via a multi case study protocol. Well-defined qualitative and quantitative MAS DSML assessment metrics will be developed. Besides, a structural guideline, which may support MAS developers while comparing and/or choosing the most appropriate DSML for the design and implementation of the intended agent system, will be produced at the end of the bilateral project.

Start Date: March 1, 2017

End Date: March 1, 2020

Total Budget: 326.980 TL (~$91,000)

Project Team:

Turkish Side:

Assoc. Prof. Dr. Geylani KARDAS (Principal Investigator)

Asst. Prof. Dr. Moharram CHALLENGER (Researcher)

Baris Tekin TEZEL (Scholar) (Ph.D. Student)

Omer Faruk ALACA (Scholar) (M.Sc. Student)

Portuguese Side:

Asst. Prof. Dr. Vasco AMARAL (Principal Investigator)

Asst. Prof. Dr. Miguel GOULÃO (Researcher)

Ankica BARISIC (Researcher)

Tomás MIRANDA (Researcher)


Supporting the Interoperability between Domain-specific Agent Modeling Languages with Horizontal Model Transformations

Research project funded by the Scientific Research Projects Directorate of Ege University under grant 16-UBE-001.

The methodology applied in the current agent domain-specific modeling language (DSML) development approaches for the derivation of operational semantics unfortunately requires the definition and implementation of new model-to-model (M2M) and model-to-text (M2T) transformations from scratch in order to make the DSMLs functional for different agent execution platforms. For each new target agent execution platform, agent DSML designers should repeat all the time-consuming and mostly troublesome steps of preparing the vertical transformations between the related DSML and this new agent platform. Instead of applying this methodology, both the definition and the implementation of M2M transformations between the metamodels of MAS DSMLs may be more convenient and less laborious in the way of enriching the support of agent DSMLs for various agent execution platforms. Hence, this project aims at improving the mechanism of constructing language semantics over the interoperability of agent DSMLs and hence providing a more efficient way of extension for the executability of modeled agent systems on various underlying agent platforms. The project also covers investigating the applicability of the proposed DSML interoperability approach by constructing horizontal transformations between full-fledged agent DSMLs.

Start Date: June 1, 2016

End Date: June 1, 2018

Total Budget: 38.950 TL (~$13,200)

Project Team:

Assoc. Prof. Dr. Geylani KARDAS (Principal Investigator)

Emine BIRCAN (Researcher) (M.Sc. Student)

Baris Tekin TEZEL (Researcher) (Ph.D. Student)

Sercan DEMIRCI (Researcher) (Ph.D. Student)

Related Publication:

Bircan, E., Challenger, M. and Kardas, G. (2016) “Interoperability of MAS DSMLs via Horizontal Model Transformations”, In proceedings of the 4th Workshop on Model Driven Approaches in System Development (MDASD 2016), held in conjunction with 2016 Federated Conference on Computer Science and Information Systems (FedCSIS 2016), September 11-14, 2016, Gdansk, Poland, IEEE Conference Publications , pp. 1555-1564.

Interoperability of agent DSMLs via horizontal model transformations.


Development of an Agent based Alternative Path Selection Method for Peer-to-Peer Video Streaming Applications

TUBITAK Academic Research Funding Program Directorate (ARDEB) Electrical, Electronics and Informatics Research Grant Committee (EEEAG) funded research project (TUBITAK ARDEB Grant no: 111E022)

One of the most challenging parts of video streaming systems running on a peer to peer network is to cope with dynamic network conditions and unpredictable node behaviors. Therefore, in an overlay network, different paths are created from source to destination and when packet transmission problems occur in one video streaming path, destination node continues receiving packets on the other path. However, it needs to be determined whether designed path diversity or selected streaming path(s) is optimal in terms of video quality and path bandwidth. A system, based on the software agents with autonomous, reactive, proactive and social features, may be suitable for solving the selection problem of alternative paths in peer to peer video streaming systems. Hence, in this project, a new path selection system based on agents was developed in order to increase service quality of nodes in a peer to peer video streaming system. Both the construction of an overlay network and the development of the internal architecture of agents were realized. Connection between the nodes in the streaming network and the software agents were set up in order to provide agent-based selection of alternative paths. Logical path selection method, developed in this project, enables nodes to infer on the capacity and robustness of the other nodes based on the reinforcement learning. Another approach brought by the project considers the assessment of the physical paths between a node and its parents. According to the simulation results, it can be said that the proposed method performs much better than the existing methods considering all criteria on both quality of experience and network conditions.

Start Date: October 1, 2011

End Date: October 1, 2013

Total Budget: 177.078 TL (~$99,000)

Project Team:

Asst. Prof. Dr. Geylani KARDAS (Principal Investigator)
Asst. Prof. Dr. Muge SAYIT (Researcher)
Cihat CETINKAYA (Scholar) (Ph.D. Student)
Sercan DEMIRCI (Scholar) (Ph.D. Student)
Yagiz KAYMAK (Scholar) (Ph.D. Student)
Kemal Deniz TEKET (Scholar) (M.Sc. Student)
Suleyman YILDIRIM (Scholar) (M.Sc. Student)

Related Publications:

1. Karayer, E. and Sayit, M. (2016) “A path selection approach with genetic algorithm for P2P video streaming systems”, Multimedia Tools and Applications, vol. 75, no. 23, pp. 16039–16057, DOI: 10.1007/s11042-015-2912-y.

2. Yildirim, S., Sayit, M. and Kardas, G. (2015) “A belief-desire-intention agent architecture for partner selection in peer-to-peer live video streaming applications”, Expert Systems, vol. 32, no. 3, pp. 327-343, DOI: 10.1111/exsy.12086.

3. Teket, K. D., Sayit, M. and Kardas, G. (2014) “Software agents for peer-to-peer video streaming”, IET Software, vol. 8, no. 4, pp. 184-192, DOI: 10.1049/iet-sen.2013.0181.

4. Karayer, E. (2014) “Inference based Alternative Path Selection for P2P Video Streaming Applications”, M.Sc. Thesis, Ege University, 95 pages (in Turkish).

5. Teket, K. D. (2013) “Agent-based Parent Selection in Peer-to-Peer Mesh-based Video Streaming Systems”, M.Sc. Thesis, Ege University, 60 pages (in Turkish).

6. Yildirim, S. (2012) “An Agent Architecture for Peer-to-Peer Live Video Streaming Applications”, M.Sc. Thesis, Ege University, 82 pages (in Turkish).

7. Sayit, M., Karayer, E., Teket, K. D., Kaymak, Y.,Cetinkaya, C., Demirci, S. and Kardas, G. (2013) "A Score-based Packet Retransmission Approach for Push-Pull P2P Streaming Systems", In proceedings of the 6th International Symposium on Multimedia Applications and Processing (MMAP 2013), held in conjunction with 2013 Federated Conference on Computer Science and Information Systems (FedCSIS2013), September 8-11, 2013, Krakow, Poland, IEEE Conference Publications, pp. 627-633.

8. Teket, K. D., Kaymak, Y., Sayit, M. and Kardas, G. (2013) "Engineering a Multi-agent System for Peer-to-Peer Video Streaming", In proceedings of 2013 IEEE International Symposium on Innovations in Intelligent Systems and Applications (INISTA 2013), June 19-21, 2013, Albena, Bulgaria, IEEE Computer Society, pp. 1-7, DOI: 10.1109/INISTA.2013.6577655.

9. Sayit, M., Kaymak, Y., Teket, K. D., Cetinkaya, C.,Demirci, S. and Kardas, G. (2013) "Parent Selection via Reinforcement Learning in Mesh-based P2P Video Streaming", In proceedings of the 10th International Conference on Information Technology: New Generations (ITNG 2013), April 15-17, 2013, Las Vegas, Nevada, USA, IEEE Conference Publications, pp. 546-551, DOI: 10.1109/ITNG.2013.89.

10. Kaymak, Y., Teket, K. D. and Sayit, M. (2013) “Parameter Analysis for a Push/Pull-based P2P Live Video Streaming Application", In proceedings of the 21st IEEE Signal Processing and Communications Applications Conference (SIU 2013), April 24-26, 2013,Girne, Northern Cyprus, IEEE Conference Publications, pp. 1-4 (in Turkish).

11. Sayit, M. and Sonmez, O. (2012) "Reinforcement Learning for Peer to Peer Video Streaming", In proceedings of the 20th IEEE Signal Processing and Communications Applications Conference (SIU 2012),Mugla, Turkey, IEEE Conference Publications, pp. 1-4 (in Turkish).

Layered system architecture

Overview of the multi-agent system developed in the project

Internal design of a peer agent

Comparison of average continuity index and startup delay

Average Continuity Index

Average Startup Delay

CoolStreaming-like System

87.7%

21.2 seconds

MAS developed in the project

92.2%

16.1 seconds


A Domain Specific Modeling Language for Semantic Web enabled Multi-agent Systems

TUBITAK Academic Research Funding Program Directorate (ARDEB) Electrical, Electronics and Informatics Research Grant Committee (EEEAG) and Slovenian Research Agency (ARRS) funded bilateral research project (TUBITAK ARDEB Grant no: 109E125)

        

Software agents are considered to be autonomous entities which contain intelligence that serves for solving their selfish or common problems, and to achieve certain goals. These agents constitute Multi-agent Systems (MAS). However, the autonomous, responsive, and proactive natures of agents make the development of agent-based software systems more complex than other software systems. Furthermore, the design and implementation of a MAS may become even more complex and difficult to implement when considering new requirements and interactions for new agent environments like the Semantic Web. Both domain-specific modeling and the use of a domain-specific modeling language (DSML) may provide the required abstraction, and hence support a more fruitful methodology for the development of MASs. Within this context, a DSML, called SEA_ML, has been developed for the design and implementation of MAS with including all of its components and supporting software tools in this project. In addition to the classical viewpoints of a MAS, the proposed DSML includes new viewpoints which specifically support the development of software agents working in the Semantic Web environment. At first, a metamodel and an abstract syntax were defined for SEA_ML. Later, both graphical and textual concrete syntaxes were developed. Upon completion of the formal definition of the semantics, operational semantics was derived via model transformations in order to provide the real implementation of the designed MAS models. Codes for the agent software can be automatically achieved as the result of applying model to code transformations. All required tools for MAS modeling and developing software according to SEA_ML were also constructed in this project.

Start Date:March 1, 2010

End Date: March 1, 2013

Total Budget: 129.144 TL (~$86,000)

Project Team:

Turkish Side:

Asst. Prof. Dr. Geylani KARDAS (Principal Investigator)

Moharram CHALLENGER (Scholar) (Ph.D. Student)

Sebla DEMIRKOL (Scholar) (M.Sc. Student)

Sinem GETIR (Scholar) (M.Sc. Student)

Slovenian Side:

Prof. Dr. Marjan MERNIK (Principal Investigator)

Dr. Tomaz KOSAR (Researcher)

Dejan HRNCIC (Researcher)

Matej CREPINSEK (Researcher)

Related Publications:

1.Challenger, M., Kardas, G. and Tekinerdogan, B. (2016) “A systematic approach to evaluating domain-specific modeling language environments for multi-agent systems”, Software Quality Journal, vol. 24, no. 3, pp. 755-795, DOI: 10.1007/s11219-015-9291-5.

2.Challenger, M., Mernik, M., Kardas, G. and Kosar, T. (2016) “Declarative specifications for the development of multi-agent systems”, Computer Standards & Interfaces, vol. 43, pp. 91-115, DOI: 10.1016/j.csi.2015.08.012.

3.Challenger, M. (2016) “A Domain-specific Modeling Language for Semantic Web enabled Multi-agent Systems”, Ph.D. Dissertation, EgeUniversity, 186 pages.

4.Getir, S., Challenger, M. and Kardas, G. (2014) “The formal semantics of a domain-specific modeling language for semantic web enabled multi-agent systems”, International Journal of Cooperative Information Systems, vol. 23, no. 3, pp. 1-53, DOI: 10.1142/S0218843014500051.

5.Challenger, M., Demirkol, S., Getir, S., Mernik, M., Kardas, G. and Kosar, T. (2014) "On the use of a domain-specific modeling language in the development of multiagent systems", Engineering Applications of Artificial Intelligence, vol. 28, pp. 111-141, DOI: 10.1016/j.engappai.2013.11.012.

6.Kardas, G. (2013) "Model-driven development of multi-agent systems: a survey and evaluation", The Knowledge Engineering Review, vol. 28, no. 4, pp. 479-503, DOI: 10.1017/S0269888913000088.

7.Demirkol, S., Challenger, M., Getir, S., Kosar, T., Kardas, G. and Mernik, M. (2013) "A DSL for the development of software agents working within a semantic web environment", Computer Science and Information Systems, vol. 10, no. 4, pp. 1525-1556, DOI: 10.2298/CSIS121105044D.

8.Getir, S. (2012) “Semantics of a Domain-specific Modeling Language for Semantic Web enabled Multi-agent Systems”, M.Sc. Thesis, Ege University, 139 pages (in Turkish).

9.Demirkol, S. (2012) “Syntax of a Domain-specific Modeling Language for Semantic Web enabled Multi-agent Systems”, M.Sc. Thesis, Ege University, 70 pages (in Turkish).

10.Demirkol, S., Challenger, M., Getir, S., Kosar, T., Kardas, G. and Mernik, M. (2012) "SEA_L: A Domain-specific Language for Semantic Web enabled Multi-agent Systems", In proceedings of the 2nd Workshop on Model Driven Approaches in System Development (MDASD 2012), held in conjunction with 2012 Federated Conference on Computer Science and Information Systems (FedCSIS 2012), September 9-12, 2012, Wroclaw, Poland, IEEE Conference Publications, pp. 1373-1380.

11.Getir, S., Challenger, M., Demirkol, S. and Kardas, G. (2012) "The Semantics of the Interaction between Agents and Web Services on the Semantic Web", In proceedings of the 7th IEEE International Workshop on Engineering Semantic Agent Systems (ESAS 2012), held in conjunction with the 36th IEEE Signature Conference on Computers, Software, and Applications (COMPSAC 2012), 16-20 July 2012, Izmir, Turkey, IEEE Computer Society, pp. 619-624, DOI: 10.1109/COMPSACW.2012.112.

12.Getir, S., Demirkol, S., Challenger, M. and Kardas, G. (2011) "The GMF-based Syntax Tool of a DSML for the Semantic Web enabled Multi-Agent Systems", In proceedings of the Workshop on Programming Systems, Languages, and Applications based on Actors, Agents, and Decentralized Control (AGERE! 2011), held at the 2nd Systems, Programming, Languages and Applications: Software for Humanity Conference (SPLASH 2011), October 23-24, 2011, Portland, USA, ACM Press, pp. 235-238, DOI: 10.1145/2095050.2095087.

13.Getir, S., Demirkol, S., Challenger, M. and Kardas, G. (2011) "Graphical Concrete Syntax of a Domain Specific Modeling Language for Semantic Web enabled Agents", In proceedings of the 5th Turkish National Software Engineering Symposium (UYMS 2011), September 26-28, 2011, Ankara, Turkey, pp. 93-100 (in Turkish).

14.Challenger, M., Getir, S., Demirkol, S. and Kardas, G. (2011) "A Domain Specific Metamodel for Semantic Web enabled Multi-agent Systems", In proceedings of the 1st International Workshop on Domain Specific Engineering (DsE@CAiSE 2011), held in conjunction with the 23rd International Conference on Advanced Information Systems Engineering (CAiSE 2011), June 20-24, 2011, London, UK, Lecture Notes in Business Information Processing, vol. 83, pp. 177-186, DOI: 10.1007/978-3-642-22056-2_19.

15.Demirkol, S., Getir, S., Challenger, M. and Kardas, G. (2011) "Development of an Agent based E-barter System", In proceedings of 2011 International Symposium on Innovations in Intelligent Systems and Applications (INISTA 2011), June 15-18, 2011, Istanbul, Turkey, IEEE Computer Society, pp. 193-198, DOI: 10.1109/INISTA.2011.5946060.

16.Kardas, G., Demirezen, Z. and Challenger, M. (2010) "Towards a DSML for Semantic Web enabled Multi-agent Systems", In proceedings of the International Workshop on Formalization of Modeling Languages (FML 2010), held in conjunction with the 24th European Conference on Object-Oriented Programming (ECOOP 2010), June 21-25, 2010, Maribor, Slovenia, ACM Press, pp. 1-5, DOI: 10.1145/1943397.1943402.

SEA_ML-based MAS development methodology

The relationships between different viewpoints of the SEA_ML’s metamodel

Agent-SWS Interaction viewpoint of SEA_ML’s abstract syntax

Graphical modeling for the agent-SWS Interaction viewpoint of a multi-agent stock exchange system in the graphical syntax tool of SEA_ML

An example of model analysis for SEA_ML instances using the Alloy tool

SEA_ML’s output performance measured after conducting a multi-case study


Assessment and Performance Measurement of Semantic Service Discovery Methods on Mobile Platforms

Research project funded by the Scientific Research Projects Directorate of Ege University under grant 2009/UBE/003.

Discovery of the most appropriate services meeting with the requirements of mobile device users is one of the crucial aims of service-oriented software systems. Although geographical closeness to the mobile user represents one of the important criteria on the determination of the services, such closeness is no more sufficient due to the diversity seen on both the requirements of the users and types of the services. Especially, improper assessment of service capability and quality causes those services to be inappropriate for the user needs despite of their geographical closeness. Research on the use of the Semantic Web and related technologies to cope with this problem recently became popular. If both the capabilities of the services and the requirements of the users can be modeled and defined according to some ontologies, matchmaking of the service capabilities and user requirements becomes more efficient and hence the suitable services will be discovered easily. Vast major of the work in the related research field covers noteworthy contributions for the semantic discovery in question. However, the performance of the proposed approaches especially on the real mobile platforms is an open issue. In order to fill this gap, performance measurement of the semantic discovery of the services, registered in various ontologies with different sizes, has been carried out in this project. Services of some eating places (e.g. cafes and restaurants), distributed over an area, are considered as the semantic services in the study. Discovery of these places according to both their geographical and semantic closeness to the mobile users has been tested inside a real implemented system.

Start Date: November 1, 2009

End Date: November 1, 2012

Total Budget: 14.000 TL (~$9,300)

Project Team:

Asst. Prof. Dr. Geylani KARDAS (Principal Investigator)

Dr. Ilker KOCABAS (Researcher)

Hamit UNUGUR (Researcher) (M.Sc. Student)

Egemen EVIN (Researcher) (M.Sc. Student)

Related Publication:

Unugur, H. (2014) “Assessment of the Semantic Service Discovery Methods on Mobile Platforms”, M.Sc. Graduation Project Report, Ege University, Turkey, 52 pages (in Turkish).

An example of listing eating places according to a semantic service search on a cellular phone. The order of listing is based on the semantic closeness of services to the user’s request.

 

An example of listing eating places, which are matched with the user request in the same semantic closeness level. At this case, services are listed according to their geographic closeness to the mobile user.

Elapsed time averages of semantic service discovery in different service ontologies with total individual numbers ranging from 100 to 20000. Each total elapsed time (given in msec) is the sum of 1) time required for sending mobile user’s search request to the server, 2) service matching performed on the server and 3) listing results on the user’s cellular phone