In this work, we focus on the OWA operator. We view this operator as t
he inner product of two vectors: a weighting vector containing the wei
ghts associated with the aggregation, and a second vector containing t
he arguments to be aggregated. We note the centrality to this operator
of the process used to index the elements in the argument vector, and
suggest some new mechanisms for ordering the arguments. In the second
part, we study the situation in which the weighting vector is context
dependent. Finally, we consider the case in which the arguments to be
aggregated are drawn from an ordinal scale; the solution to this prob
lem leads us to consider random weighting vectors. (C) 1998 John Wiley
& Sons, Inc.