Goaloriented approaches Colin Potts Georgia Tech Colin Potts

Goal-oriented approaches Colin Potts Georgia Tech © Colin Potts C 2 -1

Goals & requirements l Systems exist to meet goals or objectives » Goals are not requirements – may be ambiguous and inconsistent – not absolute (some take priority) – idealized rather than implementable – not allocated to product » Goals are not business processes – processes are implementations of goals © Colin Potts C 2 -2

Example of goal properties SATISFY patron’s info. requests inconsistent MAINTAIN library budget MAKE books available to public inconsistent idealized none allocated to Library IS © Colin Potts high priority C 2 -3

Goals, requirements and fitness for use Goals reqts. that don’t quite meet goals Sys © Colin Potts reqts. that meet goals Sys C 2 -4

Pre-requirements traceability l Pre-reqts. traceability shows where reqt. traces from Goals MAKE blah MAINTAIN user awareness of blah Reqts. 1. 1 The system shall blah, blah. . . 1. 2 If the co-worker is blah, the system shall inform the user. . . 1. 2. 1. . . thus, goals provide rationale for reqts. © Colin Potts C 2 -5

Types of goals l Achieve some desired state of affairs » MAKE <something true> » KNOW <some information> » SATISFY <some desired condition> l Maintain some desired state of affairs » KEEP <some good condition> » AVOID <some bad condition> l Improve along some trajectory » IMPROVE <some property>, etc. © Colin Potts C 2 -6

Expansion of goals MAKE meeting scheduled KNOW meeting constraints depends on © Colin Potts KNOW meeting participants’ KNOW all preferences feasible times KNOW meeting participants’ identities C 2 -7

Allocation of responsibilities to system & environment MAKE meeting scheduled KNOW meeting constraints Envt System KNOW meeting participants’ KNOW all preferences feasible times KNOW meeting participants’ identities © Colin Potts C 2 -8

Realization of goals as requirements KNOW meeting constraints allocations underlined 1. When an initiator calls a meeting, he or she will define a set of meeting constraints. 1. 1. The meeting scheduler shall display a form into which constraints may be entered. 1. 2. If any constraints are unspecified, the meeting scheduler shall insert default values. 1. 3. Meeting constraints include the following: - earliest meeting time - latest meeting time - names of invited participants © Colin Potts C 2 -9

Identifying obstacles to goals l What can go wrong to thwart a goal? » The actor responsible for MAKE goals may fail – mechanical failure or human error » The infrastructure responsible for KNOW goals may garble or misremember relevant information – communications errors – human error (forgetting, confusion) » A SATISFY goal may be intrinsically unsatisfiable – resource contention l Is it worth worrying about? » If so, add secondary requirements to defend against it or mitigate its effects © Colin Potts C 2 -10

Retracting idealized assumptions l Idealized goal & refinement KNOW schedule preferences data entry requirements © Colin Potts l Realistic goal KNOW schedule preferences data entry requirements change notification reqts. data modification requirements C 2 -11

Team exercise: Goal refinement l As a class » Define some high-level goals for the system described in the example requirements » Expand them into more detailed goals » Choose one intermediate-level goal l In teams of 2 -3 » Identify obstacles that could thwart that goal » Write a set of primary and secondary reqts. for the goal l As a class » Discuss insights achieved about the system © Colin Potts C 2 -12

Methods for goal-oriented requirements definition l Make process concrete » Define, expand, realize and “de-idealize” goals by considering concrete scenarios l Use goals as a benchmark of consensus » If people can’t agree on goals, they won’t agree on product features – Modified JAD or CRC meetings useful © Colin Potts C 2 -13

Goal-oriented approaches: how to find out more l Goal-oriented methods are just emerging from the research community » No textbooks or handbooks » Experience in telephony, electronic commerce, software reuse, BPR » Several recent action-research papers from GA Tech & Univ. Namur (Belgium) © Colin Potts C 2 -14