Response method and device, storage medium and electronic equipment

Abstract

The invention relates to a response method and device, a storage medium and electronic equipment, and the method comprises the steps: obtaining a to-be-responded target input text, and carrying out the word segmentation of the target input text, and obtaining a plurality of first target words; obtaining a pre-established semantic recognition database and a pre-trained context recognition model; determining a first to-be-determined response text of the target input text through the semantic recognition database according to the plurality of first target words; obtaining a second to-be-determined response text of the target input text through the context recognition model according to a plurality of first target words and the first to-be-determined response text; calculating a first association parameter of the first to-be-determined response text and the target input text and a second association parameter of the second to-be-determined response text and the target input text accordingto the plurality of first target words; And determining a target response text from the first response text to be determined and the second response text to be determined according to the first association parameter and the second association parameter.

Description

technical field [0001] The present disclosure relates to the field of conversational artificial intelligence, in particular, to a response method, device, storage medium and electronic equipment. Background technique [0002] In recent years, conversational artificial intelligence technology has developed rapidly. Conversational artificial intelligence means that robots can have smooth conversations with people like humans. For example, chat robots can chat with people in an open field. In terms of technical implementation It is necessary to use artificial intelligence algorithms and establish a chat response model on the basis of massive data, so that when the user inputs a sentence, the chat response model will output the corresponding answer through a series of complex calculations. [0003] At present, in the chat response method, there are mainly two methods, one is the retrieval response method, and the other is the generative response method. The results are output, ...

AI Technical Summary

Problems solved by technology

[0003] At present, in the chat response method, there are mainly two methods, one is the retrieval response method, and the other is the generative response method. The results are output, but the retrieval-based answer method is completely dependent on the corpus itself, and the output answers lack diversity; the generative answer is generally based on the recurrent neural network, through deep learning training, and the final result output is based on the model, but Answers based on the output of the generative answer method are prone to context-independent questions, and then output meaningless answers such as "I don't know" and "Okay", which will reduce the accuracy of the answer and cause poor user experience.

Images