Long Tail data or entities are those that are less commonly referenced within a knowledge graph (or any data set). As opposed to head entities, which act as hubs or central locations of interlinking, long tail entities often hold less connections to the remainder of the knowledge graph.
Additionally, the concept of tail components are often distinct from head components at both the micro and macro levels. At the micro level, the “tail” often refers to the grammatical object of a fact. Part of the creation of records for assimilation into knowledge graph entities involves the parsing of facts into head, relation, and tail components (alternatively, subject, predicate, and object components). Take the sentence “Nike bought Local Shoe Shop” as an example. Nike would be the head entity, “buyer of” could function as the relationship, and “Local Shoe Shop” would function as the tail entity.
For more info, see head entities.