180 likes | 286 Views
Goal and Requirement Change Management in Enterprise Architecture. Abelneh Teka. 13, June 2012. Background . A Computer Science master student at University of Twente, Netherlands. This research plan is based on the early works of my on going master Project.
E N D
Goal and Requirement Change Management in Enterprise Architecture Abelneh Teka 13, June 2012
Background • A Computer Science master student at University of Twente, Netherlands. • This research plan is based on the early works of my on going master Project. • Master Project: Analysis of Indirect Influence Relations. . Conducted on the Premise of BiZZdesign B.V. • 4 Supervisors Involved in this project and deserve credit: • Nelly Condori Fernandez and Ivan Kurtev: From University of Twente • Wilco Engelsman and Dick Quartel: From BiZZdesign.
Research Context: Enterprise Architecture(EA) and Goal change management. • Business environments and Processes are highly dynamic. • Technology supporting/enabling these processes is also highly dynamic. • EA designs should also be adaptable to cope with these changes to sustain the intended functionality of the EA. • Changes in business environments can be triggered by various events. • e.g. business Law changes, company policy changes, Government policy changes, technology advancements…. • But “important” changes are manifested as changes in stakeholders’ goals and requirements. • Bottom Line: Goal change management is crucial for EA adaptability. IDoPROFES:2012
Industrial Context: BiZZdesign, ArchiMate and TOGAF • BiZZdesign is a company in Netherlands involved in business process and EA designs as well as tool support and consultation for client organizations. • For EA design, it uses TOGAF and ArchiMate EA modeling language. • Both ArchiMate and TOGAF support modeling of requirements and other intentional elements like goals and stakeholders. ++ System Security Encryption Strength System Performance - - • BiZZdesign has also a tool named BiZZdesign Architect to design EA models via ArchiMate EA modeling language. IDoPROFES:2012
Industrial Problem: • The goal to goal relations are not semantically well defined in ArchiMate. • No Semantic definition ≈ No reasoning on goal Relations. • No Reasoning ≈ Can’t predict goal change impacts. • Can’t predict goal change effects ≈ Limited Adaptability of EA. • Limited Decision Support • Alternate resource allocation problem. IDoPROFES:2012
Industrial Problem: Simple Example - - - - • Should the company manger hire a tester or a marketer?
Research Questions • What are the existing formalizations of the goal oriented approaches, and what kind of reasoning do they allow? • Is the reasoning for indirect influence possible with them(after modification)? • How the previous two can be combined and extended to support analysis of alternative resource allocations? • How can we visualize these change impacts and how can we simulate this visualization? IDoPROFES:2012
What are the most relevant existing formalizations of the goal oriented approaches? • Goal and Requirement change management is not a new topic for the software engineering community. • e.g. NFR, KAOS, TROPOS, DEPRIVSIM, First order logic based definition of goal relations. • Research Question 1 • Two candidate approaches are selected/adapted: • based on Simplicity for users, applicability for indirect goal influence analysis, documentation availability and complexity for development: • Non Functional Requirement Framework • TROPOS based approach • Research Question 2. REFSQ:2012
Non Functional Reuirements Frameweork • Non functional Requirements as soft goals. • Goals can be partially or Fully satisfied as well as Denied. • Contribution relations can be Strongly positive, weakly positive as well as negative. REFSQ:2012
What kind of reasoning do these approaches allow? : Fuzzy Reasoning • NFR’s qualitative Reasoning can be extended using Fuzzy logic based inference engine. • Soft(Goal) satisfaction level measurement is difficult. • Vague nature of soft(goals). • Fuzzy logic can be applicable on these kind of vague(fuzzy) concepts. • A single element in a fuzzy set can belong to two sets. -85, Fully or Partially Denied? e.g: 10 °C, Warm or Cold temperature? REFSQ:2012
What kind of reasoning do these approaches allow? : TROPOS • TROPOS: Agent oriented software development methodology. • Supports early phases of requirement engineering activities. • Each Goal is assigned two variables: one for satisfiability and the other for deniability. • The variables can only take Full(F), Partial(P) or None (N) evidence availability for Goal Satisfiability and Deniability. IDoPROFES:2012
Preliminary Software Architecture: IDoPROFES:2012
What kind of reasoning is applicable for indirect influence management? : Preliminary Empirical Study Design • How efficient and effective is the algorithm proposed in comparison to other existing approaches? • Is the efficiency and effectiveness of the algorithm proposed affected by the functional size of “goal model”? REFSQ:2012
Current Status of the Research: Prototype application of TROPOS and fuzzy logic based NFR reasoning approaches is realized.(RQ 4) • Both approach seems to be consistent when applied to one test case study. • Performance wise, the two approaches behave similarly • TROPOS is better in handling conflicts. • Fuzzy Reasoning gives more detailed and concrete predictions. Future Work • Combining good features of both approaches (RQ3) • More test cases for validating the selected method REFSQ:2012
Do you have any question? IDoPROFES:2012
So what do we expect from the PROFES community again? • About the reasoning technique selected for goal analysis. • Possible validation approaches in addition to example test cases. IDoPROFES:2012