The initial motivation for our research was to create improved recommenders for a large
online dating site and an employment site. As we analysed the nature of these tasks,
the importance of reciprocity became apparent and we coined the term reciprocal recommender.
Given the extensive body of recommender research, we were surprised to
discover that there had been so little study of this class of recommender, even though
there are many important contexts where these recommenders can play significant
roles. Nor had there been any systematic analysis of the nature of such reciprocal recommenders,
how they differed from traditional recommenders and how this impacts
the design of their recommender algorithms.