Active Feature Acquisition via Human Interaction in Relational domains.

We consider the problem of interactive and explainable active feature elicitation in relational domains in which a small subset of data is fully observed while the rest of the data is minimally observed. The goal is to identify the most informative set of entities for whom acquiring additional relations would yield a more robust model. We assume the presence of a human expert who can interactively provide the relations. Thus there is a need for an explainable model. Consequently, we employ an relational tree-based distance metric to identify the most diverse set of relational examples (entities) to obtain more relational feature information on. The model that is learned iteratively is an interpretable and explainable model that is presented to the human expert for eliciting additional features. Our empirical evaluation demonstrates both the efficiency and the interpretability of the proposed approach.