Deep learning-based robot conversation interaction method and apparatus

Abstract

The invention provides a deep learning-based robot conversation interaction method. The method comprises the steps of receiving a conversation statement input by a user; extracting semantic feature units in the conversation statement, and performing knowledge base retrieval according to the semantic feature units to obtain a similar conversation statement; generating sentence type vectors and sentence vectors corresponding to the input conversation statement and the similar conversation statement respectively in a learning model, and outputting the similarity between a combination of the sentence vectors and the sentence type vectors for the input conversation statement and a combination of the sentence vectors and the sentence type vectors for the similar conversation statement; and selecting and outputting a reply statement corresponding to the similar conversation statement with highest similarity. According to the method, a robot can judge not only semantics according to a single word of Chinese but also statement semantics consisting of the words and the similarity among different statements composed of the same words, so that an appropriate reply is found and output more accurately for an intention of a conversation object, and the man-machine interaction experience effect is greatly improved.

Description

technical field [0001] The present invention relates to the field of intelligent robots, in particular to a deep learning-based robot dialogue interaction method and device. Background technique [0002] The current robot industry is developing rapidly, especially chat robots have been accepted by more and more people. Chatterbot (chatterbot) is a program used to simulate human dialogue or chat. The reason for chatbots is that the developers put the answers they are interested in into the database. When a question is thrown to the chatbot, it uses the similarity matching algorithm to find the most similar question from the database, and then based on the question and The corresponding relationship of the answer, give the most appropriate answer, and reply to its chat partner. [0003] Therefore, it is necessary to maintain a huge knowledge base for retrieval in the dialogue system. Whenever a user's chat sentence is received, the system will search the knowledge base for ...

AI Technical Summary

Problems solved by technology

[0005] Traditional NLP (Natural Language) technologies are generally based on methods such as TF-IDF or edit distance. These methods basically only consider the similarity between words, while ignoring the semantics after forming a sentence.

Images