Towards Humane Feedback Mechanisms in Exploratory Search

Machine learning (ML) plays a central role in modern information retrieval (IR) systems. We argue that, in IR systems for multi-session exploratory search, there are unexploited opportunities for IR document ranking models to leverage users’ knowledge about the search task to better support users’ search needs. Specifically, we propose a method to enable users to adapt an IR document ranking model according to their information needs, using an interface that supports search strategies and methods for engaging with documents known to be useful when people explore new or complex domains of knowledge. We also discuss the major challenges in creating human-centered machine learning models and interfaces for exploratory search.