CONTEXT INTERPRETATION IN NATURAL LANGUAGE PROCESSING USING PREVIOUS DIALOG ACTS

摘要

Features are disclosed for processing and interpreting natural language, such as interpretations of user utterances, in multi-turn dialog interactions. Context information regarding interpretations of user utterances and system responses to the user utterances can be maintained. Subsequent user utterances can be interpreted using the context information, rather than being interpreted without context. In some cases, interpretations of subsequent user utterances can be merged with interpretations of prior user utterances using a rule-based framework. Rules may be defined to determine which interpretations may be merged and under what circumstances they may be merged.

主权项

1. A system comprising:
a computer-readable memory storing executable instructions; and one or more processors in communication with the computer-readable memory, wherein the one or more processors are programmed by the executable instructions to at least:
obtain first audio data regarding a first utterance of a user;generate first speech processing results based at least partly on the first audio data, the first speech processing results comprising a semantic representation of the first utterance;generate a first response for presentation to the user based at least partly on the first speech processing results;store context information comprising the semantic representation of the first utterance and a semantic representation of the first response;obtain second audio data regarding a second utterance of the user;generate second speech processing results using the second audio data, the context information, and a context interpretation rule, wherein the context interpretation rule relates to replacing at least one of a slot value or an intent of the semantic representation of the first utterance with at least one of a slot value or an intent of a semantic representation of the second utterance, and wherein the context rule is based at least partly on the semantic representation of the first response; andgenerate a second response for presentation to the user based at least partly on the second speech processing results.