Thesis (M.S.)--Worcester Polytechnic Institute. / Keywords: LSR model; IFSO framework; automatic/semi-automatic software refactoring; software refactoring framework. Includes bibliographical references (p. 89-92).

Thesis (M.S.)--Bowling Green State University, 2007. / Document formatted into pages; contains ix, 60 p. Includes bibliographical references.

Thesis (M.Sc.). / Written for the School of Computer Science. Title from title page of PDF (viewed 2008/12/04). Includes bibliographical references.

Thesis (M.Sc. (Computer Science)) -- University of Pretoria, 2007. / Includes bibliographical references ( leaves 117-120)

To automatically/semi-automatically improve internal structures of a legacy system, there are several challenges: most available software analysis algorithms focus on only one particular granularity level (e.g., method level, class level) without considering possible side effects on other levels during the process; the quality of a software system cannot be judged by a single algorithm; software analysis is a time-consuming process which typically requires lengthy interactions. In this thesis, we present a framework, IFSO (Integrated Framework for automatic/semi-automatic Software refactoring and analysis), as a foundation for automatic/semi-automatic software refactoring and analysis. Our proposed conceptual model, LSR (Layered Software Representation Model), defines an abstract representation for software using a layered approach. Each layer corresponds to a granularity level. The IFSO framework, which is built upon the LSR model for component-based software, represents software at the system level, component level, class level, method level and logic unit level. Each level can be customized by different algorithms such as cohesion metrics, design heuristics, design problem detection and operations independently. Cooperating between levels together, a global view and an interactive environment for software refactoring and analysis are presented by IFSO. A prototype was implemented for evaluation of our technology. Three case studies were developed based on the prototype: three metrics, dead code removing, low coupled unit detection.

Thesis (MComm) -- University of Stellenbosch, 2001. / ENGLISH ABSTRACT: Software restructuring is a form of perfective maintenance that modifies the structure of a program's source code. lts goal is increased maintainability to better facilitate other maintenance activities, such as adding new functionality or correcting previously undetected errors. The modification of structure is achieved by applying transformations to the source code of a software system. Software engineers often attempt to restructure software by manually transforming the source code. This approach may lead to undesirable and undetectable changes in its behaviour. Ensuring that manual transformations preserve functionality during restructuring is difficult; guaranteeing it is almost impossible. One solution to the problem of manual restructuring is automation through use of a restructuring tool. The tool becomes responsible to examine each transformation and determine its impact on the software's behaviour. If a transformation preserves functionality, it may be applied to produce new source code. The tool only automates the application of transformations. The decision regarding which transformation to apply in a specific situation still resides with the maintainer. This thesis describes the design and implementation of a restructuring tool for the Oberon language, a successor of Pascal and Modula-2, under the PC Native Oberon operating system. The process of creating an adequate abstraction of a program's structure and its use to apply transformations and generate new source code are investigated. Transformations can be divided into different classes: Scoping, Syntactic, Control flow and Abstraction transformations. The restructuring tool described in this thesis contains implementations from all four classes. Informal arguments regarding the correctness of each transformation are also presented. / AFRIKAANSE OPSOMMING: Die herstrukturering van programmatuur is daarop gemik om die struktuur van 'n program se bronkode te wysig. Hierdie strukturele veranderings dien in die algemeen as voorbereiding vir meer omvangryke onderhoudsaktiwiteite, soos byvoorbeeld die toevoeging van nuwe funksionaliteit of die korrigering van foute wat voorheen verskuil was. Die verandering in struktuur word teweeggebring deur die toepassing van transformasies op die bronkode. Programmatuur-ontwikkelaars voer dikwels sulke transformasies met die hand uit. Sulke optrede kan problematies wees indien 'n transformasie die funksionaliteit, in terme van programgedrag, van die programmatuur beïnvloed. Dit is moeilik om te verseker dat bogenoemde metode funksionaliteit sal behou; om dit te waarborg is so te sê onmoontlik. 'n Oplossing vir bogenoemde probleem is die outomatisering van die herstruktureringsproses deur die gebruik van gespesialiseerde programmatuur. Hierdie programmatuur is in staat om die nodige transformasies toe te pas en terselfdertyd funksionaliteit te waarborg. Die keuse vir die toepassing van 'n spesifieke transformasie lê egter steeds by die programmeerder. Hierdie tesis bespreek die ontwerp en implementering van programmatuur om bronkode, geskryf in Oberon (die opvolger van Pascal en Modula-2), te herstruktureer. Die skep van 'n voldoende abstrakte voorstelling van bronkode, die gebruik van sodanige voorstelling in die toepassing van transformasies en die reprodusering van nuwe bronkode, word bespreek. Transformasies kan in vier breë klasse verdeel word: Bestek, Sintaks, Kontrolevloei en Abstraksie. Die programmatuur wat ontwikkel is vir hierdie tesis bevat voorbeelde uit elkeen van die voorafgenoemde klasse. Informele argumente word aangebied om die korrektheid van die onderskeie transformasies te staaf.

Abstract not available

M. Tech. Information Networks / Code clone is a code portion in one source code fragment that is similar or identical to a code portion in another source code fragment. Clones in applications are inevitable within an organization's intranet. There are a great number of clone detection tools to help maintenance programmers to locate and refactor code clones where they exist. Currently clone detection process has not been explored fully to construct digital libraries to store clones for reuse and shareability. This is because most of clone detection techniques produce Indexed Statistical Reports as textual file showing related group of code fragments. Other techniques visualize clones to depict clones versions history as genealogies. Furthermore current techniques do not indicate the reusability and shareability worthiness of the detected clones in taxonomy. In this mini-dissertation a Clone Wrapper Detection Technique prototype is developed to detect and store commonly used structural clones into a Discoverable Digital Clone Library hosted in Fedora Repository. Stored clones metadata are then extracted to induce a Clone Family Tree Ontology of related class clones in a taxonomy based on Abstraction (inheritance and composition hierarchy) process.

Thesis (M.S.E.)--University of Wisconsin -- La Crosse, 2008. / Includes bibliographical references (p. 43).

Thesis (Ph. D.)--University of Texas at Austin, 1999. / Vita. Includes bibliographical references (leaves 131-136). Available also in a digital version from Dissertation Abstracts.