In online matching, a sequence of (random) items arrives over time and groups of items are formed to maximize some performance goal. This problem exists in many forms and covers a wide range of applications, from ad placement on the web to question-and-answer pairing in forums to organ donation programs. Although online matching has first been considered mainly in an adversarial framework, with the goal of proving worst-case performance bounds, stochastic modeling and machine learning have recently approached this problem from a stochastic perspective, allowing them to obtain statistical bounds on performance and to translate existing methods from other application areas. However, despite the close links between their approaches, these two communities are still largely disjoint and have little interaction. The goal of this workshop is to bring together researchers who approach online matching from different perspectives and to foster interdisciplinary interaction.
This workshop is part of the thematic semester Stochastic control and learning for complex networks (SOLACE).