Adapting Rankers Online

At the heart of many effective approaches to the core information retrieval problem---identifying relevant content---lies the following three-fold strategy: obtaining content-based matches, inferring additional ranking criteria and constraints, and combining all of the above so as to arrive at a single ranking of retrieval units.

As retrieval systems become more complex, learning to rank approaches are being developed to automatically tune the parameters for integrating multiple ways of ranking documents. Using online learning to rank approaches, retrieval systems can learn directly from interactions with users, while they are running. Such systems can continuously adapt to user preferences throughout their lifetime, leading to better search performance in settings where expensive manual tuning is infeasible.

In the talk Maarten de Rijke focused on two issues related to online learning to rank: fist,the professor discussed the issue of balancing exploitation (that is, using what has been learned so far) and exploration (i.e., trying our alternatives so as to learn effectively). Second, presented a new method for comparing retrieval functions using implicit feedback. The method is based on a probabilistic model of such comparisons. The analytical and experimental results showed that the method is more accurate, and more robust to noise than existing methods.

The talk was based on joint work with Katja Hofmann and Shimon Whiteson.