You want R1, R2, and R3 2. Predict stakeholders’ needs (kNN)
2. Predict stakeholders’ needs (kNN) You want R1, R2, and R3 You may also want R4, R5, and R7
Predicting Requirements S.L. Lim & A. Finkelstein (2011) StakeRare: Social Networks and Collaborative Filtering for Large-Scale Requirements Elicitation. IEEE Trans. on Soft. Eng.
BUT…In StakeNet and StakeRare, Analyst interviews stakeholders
For large projects…
www.stakesource.co.uk S.L. Lim, D. Quercia & A. Finkelstein (2010) StakeSource. In 32 nd Int. Conf. on Soft. Eng.
StakeSource 2.0 S.L. Lim, D. Damian & A. Finkelstein (2011) StakeSource2.0. In 33 rd Int. Conf. on Soft. Eng.
> 50 Trial Requests Since Dec‘09
> 10 Projects Used StakeSource
IS Division, UCL
Bartlett Faculty of Built Environment , UCL
National Institute of Informatics, Japan
Project Networks Edubase (NII, Japan) IRIS (UCL ISD) RIM (British Lib) Healthy Cities (Bartlett)  S.L. Lim, F. Ishikawa, D. Damian & A. Finkelstein (2011) Using Web 2.0 to engage Stakeholders: the Good, Bad and the Ugly (submitted).  S.L. Lim & P. Bentley (2011) Evolving relationships between social networks and stakeholder involvement in software projects. GECCO’11 SBSE track .
Stakeholders Requirements Multi-Objective
Treat each set of stakeholder’s ratings as an objective to optimise
Sum of Weighted Objectives (Single Objective)
Non-Dominated Sorting (NDS)
Alternative highly multi-objective methods:
Weighted Maximum Ranking (WMR)
Weighted Average Ranking (WAR)*
Sum of Weighted Ratios (SWR)
Sum of Weighted Global Ratios (SWGR)
Pareto optimal front too large - contains many solutions that are not desirable
For ~80 objectives not feasible to visualise or select good solutions
Requirements engineers want an optimal solution, not a range of solutions to choose from