Named entity extraction in automated chat assistant

Abstract

A system for providing responses to natural language queries or commands is disclosed. The system comprises a server comprising a processor and non-transitory memory storing instructions that cause the processor to: retrieve a stored data structure comprising a plurality of named entities; receive a natural language query or command from a human user; extract one or more named entities from the query or command, wherein the extraction comprises iteratively: dividing the query or command into overlapping sets of tokens, determining whether a set of tokens corresponds to a named entity from the plurality of named entities, extracting any corresponding set of tokens from the query, and if no corresponding sets remain, decreasing the set size and repeating; and based at least in part on the query or command and on the extracted one or more named entities, act to fulfill a user intent expressed by the query or command.