Completed theses, Andreas Symeonidis, 2010

Most of the diploma theses in Greek. In each text an English abstract is provided. For further information, you may contact the primary author, or Dr. Symeonidis.

2010

 

  • Nikolaos Stasinopoulos,Algorithm of Export Semantic Knowledge From Software Storage.

    We humans are confronted with an unprecedented increase in the amount of information we intake, in virtually every part of our digital life. The field of Software Engineering, being a central foothold of the modern human activity, has been inevitably affected. As a direct consequence, software products that meet our needs have grown, evolved, multiplied. Along, software data have grown too, so much that, following the trend of the time, they could not be ignored anymore. Thus, the field of Mining Software Repositories has emerged… Related literature up to this point have been studying mainly changes in program source code, leaving out information residing in other kinds of Software Repositories, which have remained relatively unexploited. Current diploma thesis attemps to bridge the gap and exploit data coming from Software Issue Tracking Systems. Focus is given on extracting knowledge on one of the most critical facets of Software development, that of identifying bugs. We have developed a mechanism that semantically analyzes certain features of the available data, in order to detect the latent information inside the natural description coming with the bug reports. To do so, we have exploited primitives from the fields of Natural Language Processing, Semantic Analysis and Semantic Relatedness. The final product of the diploma thesis is a multi-variate mechanism for extracting semantically-aware knowledge in Software Repositories of well-known and fairly developed projects
  • Theano Mintsi, Algorithm to Export Relations to Software Requirements Using Technics of Natural Language and Data Mining.

    Up to our days, capturing Software Requirements still remains the Achilles’ heel of the whole Software Development process. The precise definition of requirements defines product success to a great extend, since errors and omissions go over the next stages of development and can cause the failure of the project. Natural Language is extensively used during the stage of eliciting Software Requirements. Although it is full of ambiguities and incomplete expressions, natural language comprises the basic way of communication between stakeholders, and is thus crucial for the development of software systems. In fact, the acquisition of domain knowledge and the elicitation of users’ needs and expectations on the final product are achieved through text analysis and interviews. Within the context of this diploma thesis a review of CASE (Computer‐aided Software Engineering) tools that support Software Requirements Elicitation and Documentation is performed. Moreover, the thesis introduces RightRequirement, a tool that aims to support the elicitation of software requirements through a linguistic approach. RightRequirement uses Natural Language Processing Techniques in order to extract knowledge, i.e. basic concepts and relations, from software requirements written in English.  The extracted conceptual model is loaded to an ontology especially designed for the needs of the current project. Finally, RightRequirement applies machine learning techniques on the individuals of the ontology in order to extract Association Rules between them and by extension between the initial requirements. RightRequirement is envisioned as the backbone of a complete Software Requirements recommendation engine to‐be‐built. Through RightRequirement a number of experiments are performed and the results prove encouraging in a small scale. Conclusions and future work are discussed at the end of the thesis.
  • Emmanouil Spanoudakis,ezHome – Simulation and Control System of a Smart House With the use of Agents.

    Nowadays the number of household appliances is constantly increasing, while they are getting equipped with more and more advanced features. This leads to an increased need for available power through the energy grid, while at the same time creating the potential for control and monitoring of the appliances within a household. During the last few years, Smart Energy Meters have begun to emerge. These instruments are capable of giving real-time measurements of power consumption to the residents of a house, as well as providing a means of communication between residents and their power supplier. In the immediate future it is expected that the widespread use of Smart Meters will help provide monitoring and control of power consumption on a household level. Households are envisioned to use less power (and spend less for power), which will also help smooth out consumption curves in groups of households. This will ultimately help prevent black-outs in large areas of the power grid. The scope of this diploma thesis is the specification, design and implementation of a software platform, which will provide the residents of a house full control over their electric appliances, information on their power consumption both in terms of the amount of KWh consumed as well as on the cost of their consumption. The platform will also work as an intermediate between the house and the power supplier. ezHome, which has been developed within the context of the diploma thesis, is based on Software Agents and provides all the aforementioned functionality, as well as the ability to schedule the operation of any device at the optimal cost-effective time slot, based on previous days KWh prices. All the operations can be performed by the user through the system’s friendly Graphical User Interface.
  • Iraklis Tsekourakis, Trust and Famous Management in Untrusted Stystems of Software Agents.