Ranking Related Entities: Components and Analyses (Abstract)

Related entity finding is the task of returning a ranked list of homepages of relevant entities of a specified type that need to engage in a given relationship with a given source entity. We propose a framework for addressing this task and perform a detailed analysis of four core components; co-occurrence models, type filtering, context modeling, and homepage finding. Results show that pure co-occurrence is useful to select initial candidates, that type filtering is an instrument for tuning towards either recall or precision, and that context models successfully promote entities engaged in the right relation with the source entity. Our method achieves very high recall scores on the end-to-end task and is able to incorporate additional heuristics that lead to state-of-the-art performance.