Hurevren Kilic

Measurement theory and validation criteria for software complexity measures

The relation between measurement theory and evaluating criteria for software complexity measure is well established by several researchers. A new software complexity measure should also be satisfied by measurement theory criteria. However, most of the developers of new complexity measures do not care about measurement theory. In this paper, we evaluate measurement theory and point out why measurement theory is not well applicable for evaluating complexity measure.

An Empirical Study About Search-Based Refactoring Using Alternative Multiple and Population-Based Search Techniques

Automated maintenance of object-oriented software system designs via refactoring is a performance demanding combinatorial optimization problem. In this study, we made an empirical comparative study to see the performances of alternative search algorithms under a quality model defined by an aggregated software fitness metric. We handled 20 different refactoring actions that realize searches on design landscape defined by combination of 24 object-oriented software metrics. The investigated algorithms include random, steepest descent, multiple first descent, multiple steepest descent, simulated annealing and artificial bee colony searches. The study is realized by using a tool called A-CMA developed in Java that accepts bytecode compiled Java codes as its input. The empiricial study showed that multiple steepest descent and population-based artificial bee colony algorithms are two most suitable approaches for the efficient solution of the search based refactoring problem.

Process Matchmaking on a P2P Environment

A process matchmaking environment based on P2P architecture and Gnutella protocol is established. Java agent development framework (JADE) is used as middleware. The processes are modeled as one-input transition systems augmented by goal state descriptions. A polynomial-time algorithm for handling matchmaking of peer process encounters is developed. The environment can easily be customized to a specific application domain by simple user-interface modifications and through the development of related state ontologies

An Investigation about Process Matchmaking Performances of Unstructured and Decentralized Digital Environments

Efficient matchmaking is an important problem in unstructured and decentralized digital environments. We report the results of simulations of these environments and provide measurements of match performance focusing on unstructured and decentralized environments. In the study, different environment models including P2P (Hypergrid, Gnutella with/without ultrapeer); small-world (Watts-Strogatz); heavy-tailed and random versions of Autonomous System Waxman model showing power-law distribution property; random and 2D-Grid are considered. The flooding mechanism enabling process encounters for match purposes is uninformed breadth-first-search. Simulations show that the matchmaking performance of random environment outperforms the others for almost all different problems and time-to-live settings. On the other hand, the total cost of small world environment model is the highest for almost all setups.

Mobile Prisoner's Dilemma game played on diverse habitats

Promotion of cooperative behavior in Prisoners Dilemma (PD) game while players that are allowed to move between different gaming environments (i.e. habitats) is investigated. The stochastic mobile model under study is realized over connected habitats that are situated on a two dimensional grid environment. The players appearing in the same habitat are allowed to interact with their immediate neighbors. Mobility of a player is defined as movement from its habitat to another based on both obtained payoff and randomly assigned habitat diversity values. By the end of extensive experimentation, it is concluded that player mobility is an effective factor that contributes to promotion of cooperation in spatial evolutionary games. Also, even for higher values of temptation of PD game, habitat diversity supports and triggers a collective resistance for the emergence and promotion of cooperative system behavior.

A local behavior identification algorithm for generative network automata configurations

Relation between the part and the whole is investigated in the context of complex discrete dynamical systems. For that purpose, an algorithm for local behavior identification from global data described as Generative Network Automata model configurations is developed. It is shown that one can devise a procedure to simulate finite GNA configurations via Automata Networks having static rule-space setting. In practice, the algorithm provides an automated approach to model construction and it can suitably be used in GNA based system modeling effort.

Peer-Con: A process contracting overlay

A peer to peer process contracting overlay what we call Peer-Con is developed. The system is an extension of Java agent development framework (JADE) and uses IEEE Foundation for Intelligent Physical Agents (FIPA) agent communication language (ACL) standard. Cost aware flexible representation of process capabilities; description of an operator to decide on whether given capabilities turnout to an agreement or not and self organization of peer connectivity for better contracting performance are distinguishing features of the system. The system can easily be adapted to different domains while the core functionality remains the same. Practical use of Peer-Con is shown by an application from digital signal processing module (DSPM) product planning domain.