Intelligent learning method of text customer service robot based on big data

Abstract

In order to improve the intelligence and accuracy of man-machine dialogue, the present invention provides a method for intelligent learning of a text customer service robot based on big data, including: (A) utilizing the dialogue big data O in the form of text for training to perform according to the context Classification; (B) Detect favorability information Ginfo, similar sentence repetition information Iinfo, dialogue duration information Linfo, and train big data O. The calculation process of the invention is fast, and the self-learning efficiency after SVM training is greatly improved.

Description

technical field [0001] The present invention relates to the technical field of artificial intelligence, and more specifically, to an intelligent learning method for a text customer service robot based on big data. Background technique [0002] In the current man-machine dialogue question answering system, when the user enters a question, identifying the user's intention to ask is the core part of the entire question answering system. The intention recognition is correct but the accuracy rate is too low, which will cause the answer to be returned to the user later. There are so many questions that it is impossible to choose the best answer; the wrong intention recognition will cause the inability to understand the meaning of the user, thus providing the user with an unwanted answer or being unable to give the answer directly. Existing question answering systems are mainly realized through computer algorithmic logic, and the basic process includes three processes of question a...

AI Technical Summary

Problems solved by technology

[0003] Existing intent recognition methods are based on manually labeling a large amount of corpus for training and prediction. Due to the need for a large amount of manual labeling, there will be many uncontrollable factors. For example, each labeler’s understanding of the corpus will lead to different labels. As a result, there will be repeated annotation results for the same question, and annotations of the same corpus in different classification annotations; and when it is necessary to add a new intent classification, it must be discussed and determined by relevant personnel, and then the annotation staff can be trained. Start labeling work, the machine cannot automatically add new categories

A lot of manpower and material resources will be consumed in the whole process of training the model, and many uncontrollable factors will affect the speed and progress of function training