Results 1 
5 of
5
Increasing the Semantic Similarity of ObjectOriented Domain Models by Performing Behavioral Analysis First
, 2006
"... The main goal of any objectoriented analysis (OOA) method is to clarify a problem by modeling the problem and its domain. Therefore, the most important artifact that results from OOA is the domain model, which is usually realized as a class diagram that describes the core concepts in the domain and ..."
Abstract

Cited by 4 (2 self)
 Add to MetaCart
(Show Context)
The main goal of any objectoriented analysis (OOA) method is to clarify a problem by modeling the problem and its domain. Therefore, the most important artifact that results from OOA is the domain model, which is usually realized as a class diagram that describes the core concepts in the domain and their relationships. Ideally, a mature engineering process is repeatable: analysts given the same problem and instructions to follow the same OOA process should produce semantically similar domain models. This work compares the observed semantic similarity among the different domain models produced by one process for one system by different users of the process when the process is one of: 1. creation of use cases (UCs), then sequence diagrams, then a domain model, and 2. creation of UCs, then a unified UC statechart, then a domain model. One process was used to produce 31 specifications of a large VoIP system and its accompanying information management system. The other process was used to produce 34 specifications of the same system. The data show that domain models produced using the second process were 10% more semantically similar to each other than those produced using the first process, but at a cost, by one measure, of up to 25 % more time, spent in learning the process and in requirements elicitation. 1
On the Predictability of ObjectOriented Analysis Models: Semantic Similarity and Completeness Analysis
"... Ideally, the results of a mature engineering process are predictable, i.e., independent applications of the process produce results that are the same within certain controlled limits. We analyze the predictability of domain models specified using two objectoriented analysis methods. We measure pred ..."
Abstract
 Add to MetaCart
(Show Context)
Ideally, the results of a mature engineering process are predictable, i.e., independent applications of the process produce results that are the same within certain controlled limits. We analyze the predictability of domain models specified using two objectoriented analysis methods. We measure predictability by comparing how semantically similar and complete are the domain models produced by independent analysts given the same problem. Our results are based on evaluation of 65 independently specified domain models of a large VoiceoverIP system and its supporting information management system. 1
A Possible Metric for Semantic Similarity of ObjectOriented Analysis Models and Its Use to Estimate the Predictability of ObjectOriented Analysis Methods
"... Ideally, the results of a mature engineering process are predictable, i.e., independent applications of the process produce results that are the same within certain controlled limits. We suggest estimating the predictability of domain models produced using an objectoriented analysis method by compa ..."
Abstract
 Add to MetaCart
(Show Context)
Ideally, the results of a mature engineering process are predictable, i.e., independent applications of the process produce results that are the same within certain controlled limits. We suggest estimating the predictability of domain models produced using an objectoriented analysis method by comparing how semantically similar and complete are the domain models produced by independent analysts applying the method to the same problem. We introduce a possible measure of semantic similarity and evaluate it by using it to compare the predictability of two objectoriented analysis methods applied to 65 independently specified domain models of a large VoiceoverIP system and its supporting information management system. 1.
Improving the Semantic Similarity of ObjectOriented Domain Models
"... The main goal of any objectoriented analysis (OOA) method is to clarify a problem by explicitly modeling both the problem and its domain. Therefore, the most important artifact that results from OOA is the domain model, which is usually realized as a class diagram that describes the core concepts i ..."
Abstract
 Add to MetaCart
(Show Context)
The main goal of any objectoriented analysis (OOA) method is to clarify a problem by explicitly modeling both the problem and its domain. Therefore, the most important artifact that results from OOA is the domain model, which is usually realized as a class diagram that describes the core concepts in the domain and their relationships. Ideally, a mature engineering process is repeatable: analysts given the same problem and instructions to follow the same OOA process should produce semantically similar domain models. This work compares the observed semantic similarity among the different domain models produced by one process for one system by different users of the process when the process is one of: 1. creation of use cases (UCs), then sequence diagrams, then a domain model, and 2. creation of UCs, then a unified UC statechart, then a domain model. One process was used to produce 31 specifications of a large VoIP system and its accompanying information management system. The other process was used to produce 34 specifications of the same system. The data show that domain models produced using the second process were 10% more semantically similar to each other than those produced using the first process, but at a cost, by one measure, of up to 25 % more time, spent in learning the process and in requirements elicitation. 1