FUSE: Multi-faceted Set Expansion by Coherent Clustering of Skip-Grams

Set expansion aims to expand a small set of seed entities into complete relevant entities. Most existing approaches assume the input is unambiguous and completely ignore multi-faceted semantics As result, given {"Canon", "Sony", "Nikon"}, previous models return one mixed that are either Camera Brands or Japanese Companies. In this paper, we study task expansion, which capture all semantic facets in multiple sets entities, for each facet. We propose an unsupervised framework, FUSE, consists three major components: (1) facet discovery module: identifies entity by extracting clustering its skip-grams, (2) fusion discovers shared entire optimization formulation, (3) expands utilizing masked language model with pre-trained BERT models. Extensive experiments demonstrate FUSE can accurately identify generate quality