L. C. Briand, J. W. Daly, and J. W ust. A unified framework for cohesion measurement in object-oriented systems. Empirical Software Engineering: An International Journal, 3(1):65--117, 1998.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....flattening process every client class with the possibility of calling operations of any subclass, is now also coupled with it. Although it is very straight forward to consider flattening in measurement it had not been yet examined in detail, but nearly mentioned as a theoretic possibility (cf. [1], Section 4.2.4) or treated by various rare used metrics without a substantial knowledge of their behaviour or implementation details (e.g. number of methods overridden by a subclass, number of methods inherited by a subclass, or number of methods added by a subclass, cf. 9] One point might be ....

Briand, L.C., Daly, J.W., Wst, J.: A unified framework for cohesion measurement in object-oriented systems. Fraunhofer Institute for Experimental Software Engineering, Technischer Bericht ISERN-97-05 (1997)

....and diligently making precise choices at each stage of the composition. Furthermore, by building this approach on an existing denotational semantics for object oriented programming we continue to be able to process the same system at an extremely low level. Cohesion and coupling analysis[6, 19, 21, 22, 36], slice metrics production[23, 32] and other traditional code analysis tech Component ConcreteDecoratorB addedBehaviour( ConcreteDecoratorA ConcreteComponent component component.operation( Decorator: operation( addedBehaviour( ExtendMethod:OriginalBehaviour ....

....= G) G) Conglomeration[1] b) Super (A : B, or a = super) i. Same (F = G) ExtendMethod[3] G) RevertMethod[2] i. Same (F = G) Redirect[6] Child is another possibility here, and a call to Same maps to BETA s inner, for example. The notation is taken from Abadi and Cardelli s sigma calculus [1] A : B reads A is a subtype of B Child is possible here as an addition as well, although we do not do so at this time. Obj Type ....

[Article contains additional citation context not shown here]

L.C. Briand and J.W. Daly. A unified framework for cohesion measurement in object-oriented systems. In Proc. of the Fourth Conf. on METRICS'97, pages 43--53, November 1997.

....Specifically, ISO I EC 9126:1991 [64] defines a software quality metric as a quantitative scale and method which can be used to determine the value a feature takes for a specific software product . 2 Theoretical validations of many of the metrics that we consider in this paper can be found in [20][21] 30] Early prediction is commonly cast as a binary classification problem. 3 This is achieved through a quality model that classifies components into either a high or low risk category. The definition of a high risk component varies depending on the context of the study. For example, a high ....

L. Briand, J. Daly, and J. Wuest: "A Unified Framework for Cohesion Measurement in ObjectOriented Systems". In Empirical Software Engineering, 3:65-117, 1998.

....on the definition of the metrics, and may not reflect their actual behavior. Coupling metrics characterize the static usage dependencies amongst the classes in an object oriented system [21] Cohesion metrics characterize the extent to which the methods and attributes of a class belong together [16]. Inheritance metrics characterize the structure of the inheritance hierarchy. Complexity metrics, as used here, are adaptations of traditional procedural paradigm complexity metrics to the object oriented paradigm. Current methodological approaches for the validation of object oriented product ....

L. Briand, J. Daly, and J. Wuest: "A Unified Framework for Cohesion Measurement in ObjectOriented Systems". In Empirical Software Engineering, 3:65-117, 1998.

....Specifically, ISO IEC 9126:1991 [63] defines a software quality metric as a quantitative scale and method which can be used to determine the value a feature takes for a specific software product . Theoretical validations of many of the metrics that we consider in this paper can be found in [20][21] 30] Early prediction is commonly cast as a binary classification problem. This is achieved through a quality model that classifies components into either a high or low risk category. The definition of a high risk component varies depending on the context of the study. For example, a high ....

L. Briand, J. Daly, and J. Wuest: "A Unified Framework for Cohesion Measurement in ObjectOriented Systems". In Empirical Software Engineering, 3:65-117, 1998.

..... ExtendMethod[3] G) RevertMethod[2] i. Same (F = G) Redirect[6] G) Delegate[6] i. Same (F = G) RedirectInFamily[4] i. Same (F = G) RedirectInLimitedFamily[5] Each of these captures a concept as much as a syntax, as we originally intended. Each expresses a direct and explicit way to solve a common problem, providing a structural guide as well as a conceptual abstraction. In this way they fulfill the requirements of a pattern, as generally defined, ....

....diligently making precise choices at each stage of the composition. Furthermore, by building this approach on an existing denotational semantics for object oriented programming, # calculus, we continue to be able to process the same system at an extremely low level. Cohesion and coupling analysis[5, 16, 18, 19, 32], slice metrics production[20, 28, 29] and other traditional code analysis techniques[8, 10, 30] are still completely possible within the greater # # # calculus. We have provided the link between patterns, as conceptual entity descriptions, to the formal semantics required and used by compilers ....

L. Briand and J. Daly. A unified framework for cohesion measurement in object-oriented systems. In Proc. of the Fourth Conf. on METRICS'97, pages 43--53, Nov. 1997. Albequerque.

....between 0 and 1. Although their theoretical validation is not fully convincing, the above measures (except LCOM) have been found generally useful in building models for various attributes. Many other measures have been defined. A comprehensive set is proposed in [21] Surveys can be found in [26, 27]. 7.4.5. Functional Cohesion Three measures for functional cohesion are introduced in [28] based on the data tokens (i.e. the occurrences of a definition or use of a variable or a constant) that appear in a procedure, function, or main program. Each data token is on at least one data slice, ....

L. Briand, J. Daly and J. Wuest, A Unified Framework for Cohesion Measurement in ObjectOriented Systems," Empirical Software Engineering - An International Journal(1998).

....quantify properties of individual classes and their relationships to the rest of a system. Chidamber and Kemerer s suite of object oriented measures include only single class measures [17] Other researchers, including one of the authors, have developed cohesion measures for individual classes [5, 11, 31, 39]. Work on quantifying interclass properties (properties of collections of classes) includes Hitz and Montazeri s work to quantify coupling in object oriented systems [31] They classify individual dependencies between classes and between objects, and they propose an ordinal measure of the ....

L. Briand, J. Daly, and J. Wust. A unified framework for cohesion measurement in object-oriented systems. Empirical Software Engineering, 3(1):65--117, 1998.

....a class may play a role in a design pattern and or an inheritance hierarchy. This role represents its architectural context. Most currently used OO design measures quantify properties of lower level design units such as classes, attributes, methods etc. to evaluate the designs. Briand et al. [2, 4, 3, 5] studied coupling and cohesion measures that quantify OO software design quality. Their investigations of coupling measures do not directly address the coupling between classes that represent specific roles within architectural contexts such as design patterns [4] Chidamber and Kemerer s OO ....

L. Briand, J. Daly, and J. Wust. A unified framework for cohesion measurement in object-oriented systems. Empirical Software Engineering, 3(1):65--117, 1998.

....i , m ) is undefined 0 iff dist(m i , m j ) 1 Now we are able to reproduce the original LCOM measure as follows: Additionally Chidamber and Kemerer limit this measure to a minimum of 0. Many criticisms that covered this measure can now be formulated in a more formal way (cf. e.g. Hend96] or [BrDaW98] for some LCOM improvement collections) Because LCOM only considers the properties of using some attributes the distance measure holds only for a very specific point of view, i.e. similarity through direct common attribute usage. It may happen that LCOM uses a similarity incompatible set of ....

....to encapsulation and inheritance structure 4 System level The whole system Accepted for FESMA99, Amsterdam 4. 8. October 7 18 Additionally we introduce a new intermediate level between object level and system level, that is called subsystem level (cf. cluster in [Meye97] class set in [BrDaW98] or subsystem in [Rpi94] and[Hend96] Level nr. Level name Considered entity family 3.5 Subsystem level Collection of abstract data types which are somehow related. Here we follow our observation, that large software systems heavily use subsystems as a higher abstraction than classes, and ....

[Article contains additional citation context not shown here]

Lionel C. Briand, John W. Daly, Jrgen Wst: "A unified framework for cohesion measurement in object oriented systems", Empirical Software Eng.., an Int'l J., vol.3, no.1, pp.65-117, 1998

.... Briand et al. introduce cohesion and coupling measures for object based systems based on interactions between declarations (Briand, Morasca and Basili, 1993) In related work, Briand et al. develop a framework for developing and analyzing both cohesion and coupling in object oriented systems (Briand, Daly and Wust, 1997 and 1978; Briand, Devanbu and Melo, 1997) 8.2. Software Restructuring Slicing and Restructuring. The restructuring method of Kim, Kwon, and Chung (1994) makes use of a notion of module strength (cohesion) They define processing blocks which are similar to the data slices of Bieman and Ott ....

Briand, L.; Daly, J. and Wust, J. (1997) `A unified framework for cohesion measurement in objectoriented systems', in Proceedings of the Fourth International Software Metrics Symposium, IEEE Computer Society Press, Los Alamitos CA, pp. 43--53.

....may have been designed inappropriately. Bieman95] Chapter 3 is devoted to a formal validation of the Bieman Kang cohesion measures. 2.3. 6 Taxonomy of Briand Briand et al. review the cohesion measures currently under proposal with the intent of unifying them under a common framework [Briand97a][Briand97b]. A unified empirical relation system is proposed with the following properties, informally stated: ffl The Nonnegativity and Normalization property says that cohesion values must lie in a bounded interval. ffl Null Value and Maximum Value says that cohesion is zero if there are no relationships ....

....classes must not result in a cohesion value for the merged class which is greater than those of its constituent classes. The Bieman Kang cohesion measures are shown to fulfill this empirical relation system. The authors make some interesting observations about cohesion measures in general [Briand97b]: ffl Indirect connections appear to be better indicators than direct connections for showing whether a class should be split. ffl A class having maximum cohesion as measured by direct connections would be required to possess direct connections between every element to every other element. This ....

Lionel Briand, John Daly, Jurgen Wust. "A Unified Framework for Cohesion Measurement in Object-Oriented Systems." Proceedings of the International Software Metrics Symposium (upcoming).

....prediction models. In addition, the success and widespread diffusion of object oriented software systems have drawn a good deal of interest towards the study of the attributes of objectoriented software systems. A number of studies have been published (see for instance [19] 8] 28] and [5] [6] for an extensive survey) These studies generally deal with the proposal of new measures or the reuse of existing ones in the framework of object oriented software code. Our study goes one step in the direction of object orientation, at the high level design stage, in that it addresses ....

....variable instances is larger than that of the ones not sharing any instances. Several other measures have been proposed in the literature for object oriented cohesion, e.g. see [8] 28] Due to space constraints, no thorough comparison can be made here. The interested reader is referred to [6], where an extensive survey and classification have been proposed. From a general perspective, these measures differ according to their underlying experimental hypotheses and properties. At a higher level, several criteria capture the main differences: the types of connections dependencies that ....

L.C. Briand, J. Daly, and J. Wuest, "A Unified Framework for Cohesion Measurement in Object-Oriented Systems," Empirical Soft- ware Eng.: An Int'l J., vol. 3, no. 1, pp. 65--117, 1998. Also available at http://www. iese.fhg.de/ISERN/pub/isern.biblio.html.

....Note that there are no measures used to capture class cohesion. This is because the information required by existing measures is not usually available at high level design, e.g. method attribute interactions, method method interactions, and pairs of methods which reference common attributes [6]. None of this information was present in the system designs, so cohesion cannot be compared across systems in a quantifiable manner. However, it should be pointed out that since cohesion and coupling are expected to be interdependent, this is not considered a serious issue in the context of this ....

....want to ensure the comparability of experimental materials such as requirements or design descriptions. The improvement of software encjineerincj experiments and product measurement therefore goes hand in hand. Some of our related work on the topic of object oriented measurement can be found in [3][6][27] 28] 29] 30 A replication package [9] is available for researchers interested in externally replicating our experiment. Improvements to the experimental procedure might include increasing the task time and improving the time data collection proce dures. Acknowledgments ....

L. Briand, J. Daly, and J. Wuest. A unified framework for cohesion measurement in object-oriented systems. Empirical Software Engineering, An International Journal, 3(1 ), 1998.

....can be found [31] This is, however, a pure mathematical artifact, which makes any analysis of the impact of size on fault proneness difficult. 2. 3 Independent variables The measures of coupling, cohesion, and inheritance identified in a literature survey on object oriented design measures [7][8] are the independent variables used in this study. We focus on design measurement since we want the measurement based models investigated in this paper to be usable at early stages of software development. Furthermore, we only use measures defined at the class level since this is also the ....

....The significant measures LCOM1, LCOM2, and LCOM3 show positive correlation coefficients. This indicates that the higher the values of these measures, the more fault prone the class is likely to be. This is consistent with our Hypothesis H Coh that low cohesion is bad design. As was discussed in [8], the mathematical properties of ICH make it unlikely to be measuring cohesion. ICH possesses properties of a complexity measure. With this information, the positive coefficient of ICH is reasonable: the higher ICH (and thus class complexity) the more fault prone the class. For LCC, the ....

[Article contains additional citation context not shown here]

L. Briand, J. Daly, J. Wst, "A Unified Framework for Cohesion Measurement in Object-Oriented Systems", Empirical Software Engineering Journal, 3 (1), 65-117, 1998. Also available as Technical Report ISERN-97-05 .

No context found.

Briand, L., Daly, J., Wst, J., 1998a. A Unified Framework for Cohesion Measurement in Object-Oriented Systems. Empirical Software Engineering Journal, 3 (1), 65-117.

....instances is larger than that of the ones not sharing any instances. Several other measures have been proposed in the literature for object oriented cohesion (e.g. see [CK94, BK95, HM95] Due to space constraints, no thorough comparison can be made here. The interested reader is referred to [BDW97], where an extensive survey and classification have been proposed. From a general perspective, these 21 measures differ according to their underlying experimental hypotheses and properties. At a higher level, several criteria capture the main differences: the types of connections dependencies ....

L. Briand, J. Daly, J. Wuest, "A Unified Framework for Cohesion Measurement in Object-Oriented Systems", to be published in Empirical Software Engineering: An International Journal, also available at www: http://www.iese.fhg.de/ISERN/pub/isern.biblio.html. 41

....predictive models. The results of this analysis are then presented in Section 4.0. We draw our conclusions in Section 5.0. 2.0 The Empirical Study 2. 1 Dependent Variable The goal of this study was to empirically assess the object oriented design measures discussed in a literature review [BDW96, BDW97] More specifically, we wanted to evaluate whether these measures are useful for predicting the probability of detecting faulty classes. Assuming testing was performed properly and thoroughly, the probability of fault detection in a class 1 during acceptance testing should be a good indicator ....

....was therefore the most straightforward and practical measure of fault proneness we could use, and a suitable dependent variable for our study. 2. 2 Independent Variables The measures of coupling, cohesion and inheritance identified in a literature survey on object oriented design measures [BDW96, BDW97] are the independent variables used in this study. More specifically, we focus on measures defined at the class level, although there also exist measures defined for attributes, methods, and (sub)systems. Measures at the method and attribute level could not be validated, as the fault data we have ....

[Article contains additional citation context not shown here]

L. Briand, J. Daly, J. Wüst, "A Unified Framework for Cohesion Measurement in Object-Oriented Systems", Empirical Software Engineering Journal, 3 (1), 65-117, 1998. Also available as Technical Report ISERN-97-05.

No context found.

L. Briand, J. Daly, J. Wüst, "A Unified Framework for Cohesion Measurement in ObjectOriented Systems", Empirical Software Engineering Vol 3, No. 1, 65-117, 1998.

No context found.

L. C. Briand, J. W. Daly, and J. W ust. A unified framework for cohesion measurement in object-oriented systems. Empirical Software Engineering: An International Journal, 3(1):65--117, 1998.

No context found.

L. Briand, J. Daly, J. Wuerst. A Unified Framework for Cohesion Measurement in ObjectOriented Systems. Fraunhofer Institute for Experimental Software Engineering, Germany, 1997 -- ISERN-97-05

No context found.

L. C. Briand, J. W. Daly, and J. W ust. A unified framework for cohesion measurement in object-oriented systems. Empirical Software Engineering: An International Journal, 3(1):65--117, 1998.

No context found.

L. Briand, J. Daly, and J. Wuest: "A Unified Framework for Cohesion Measurement in ObjectOriented Systems". In Empirical Software Engineering, 3:65-117, 1998.

No context found.

L. C. Briand, J. W. Daly, and J. Wust. A unified framework for cohesion measurement in object-oriented systems. Empirical Software Engineering: An International Journal, 3(1):65--117, 1998.

No context found.

Briand, L.; Daly, J. and Wust, J. (1998), `A unified framework for cohesion measurement in object-oriented systems', Empirical Software Engineering, 3(1), 65--117.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC