Automatic generation method and system for replies of chatting robot

Abstract

The invention relates to an automatic generation method and system for replies of a chatting robot. The method comprises the steps that key words in a statement input by a user are extracted, searching is conducted with the key words as indexes, and the corresponding question-answer pair is found in a prior knowledge base; if the corresponding question-answer pair is found, the replay sentence inthe question-answer pair is taken as a reply statement to be input; if the corresponding question-answer pair is not found, the statement input by the user is input an answer generation module, and areplay statement is obtained and output. By means of the automatic generation method and system for the replies of the chatting robot, the conversations automatically generated by the robot can be closer to the conversations between real people, and the automatic generation method and system can be used in a development field.

Description

technical field [0001] The invention relates to the technical field of intelligent robots, in particular to a method and system for automatically generating replies of chat robots. Background technique [0002] The human-computer dialogue technology in the open field needs to output responses that conform to the current dialogue scene and context according to user input. At present, there are many ways to generate automatic robot replies, which can be roughly divided into the following two categories: [0003] (1) Retrieval-based reply generation method: store <input S, output R> question-answer pair in the database, for a new input user request, query the similarity between "input S" in the question-answer pair in the database and the newly input sentence, and The "output R" corresponding to the "input S" with the highest similarity is the robot's response to the user input. The disadvantage of this method is that it requires a lot of effort to build a database, and...

AI Technical Summary

Problems solved by technology

The disadvantage of this technique is that the maximum relief estimation objective function used in the process of training the model is too simple and cannot fully capture all the key points that define an ideal dialogue, resulting in a dialogue that is too rigid, repetitive and generic, and lacks the understanding of the above in-depth understanding of

In other words, the replies generated by this method are difficult to guarantee the consistency of grammatical correctness and context, etc., and the generated dialogues are less anthropomorphic, unlike real dialogues between people.

[0005] To sum up, the existing human-computer dialogue technology cannot generate automatic and coherent reply sentences, and cannot be applied to the open field. At present, there is an urgent need for a chat robot that can be applied to the open field and can conduct coherent dialogue

Images