Method and device for correcting natural language generation results

Abstract

The embodiment of the invention provides a method and device for correcting a natural language generation result. The method comprises the steps: firstly, obtaining a first template sentence to be corrected in the round of iteration, wherein the sentence is generated for a dialogue action, and the dialogue action comprises a plurality of word slots; and inputting the first template sentence into apointer rewriter, and sequentially determining the characters of each output position in the sentence to be output one by one by the pointer rewriter, thereby generating a corrected second template sentence; performing word slot extraction on the second template sentence, and judging whether the extracted word slots are consistent with a plurality of word slots or not; and if not, determining thesecond template sentence as the input of the next round of iteration, generating a training sample based on the dialogue action and the second template sentence, and adding the training sample into asample buffer. The sample buffer is used for collecting training samples, and the collected training samples are used for training the pointer rewriter.

Description

technical field [0001] One or more embodiments of this specification relate to the field of machine learning, in particular to natural language generation, and methods and devices for modifying natural language generation results. Background technique [0002] With the rapid development of artificial intelligence and machine learning, various natural language processing tasks have been widely used in various business implementation scenarios. In various business scenarios such as intelligent customer service and automatic broadcasting, the automatic dialogue system will be used to generate natural language dialogue sentences. [0003] The core component of the automatic dialogue system is the natural language generation NLG (Natural Language Generation) system, which is used to convert a meaning representation, that is, a dialogue action DA (Dialogue Act) into a natural language sentence. Generally, a dialogue action DA contains an action type (act type) and some word slot-...

AI Technical Summary

Problems solved by technology

However, existing NLG systems still often have conversion errors, for example, missing some word slots, or misplaced word slots, etc., and the generated sentences are often not natural and smooth enough.

Images