A Variety Metric Accounting for Unbalanced Idea Space Distributions

Proving the effectiveness of an idea generation method is key to its acceptance in an industrial and academic environment. This necessitates the development of a set of widely accepted metrics covering the different aspects on which idea generation methods can be characterized. This paper gives an overview of the existing metrics, and demonstrates a number of shortcomings in the variety metric, such as not accounting for the fairness of the distribution of ideas over nodes on an abstraction level. A level-based, correctly normalized variety metric, based on the Shannon entropy, is proposed which is shown to resolve the identified issues.