Text classification method, question-answering system and dialogue robot

Abstract

The invention discloses a text classification method, a question-answering system and a dialogue robot, and the method comprises the steps: predicting data of a verification set to acquire an initialcategory, a prediction category and a prediction score of the verification set; selecting a category I as a category to be optimized, and extracting a data set A from the verification set according tothe category I to be optimized; extracting a data set Ai from the data set A, wherein the initial category of the data set Ai is a category Ii; sorting the data set Ai according to the prediction score and calculating a statistical score Si; and finally, inputting the to-be-classified text into the target model for prediction, and outputting a final prediction category, thereby realizing a methodfor improving the recognition accuracy of the specific category under the condition of not changing the data model, and saving the labor and time cost of data supplement and model adjustment.

Description

technical field [0001] The invention relates to the technical field of artificial intelligence, in particular to a text classification method and a question answering system and a dialogue robot applying the method. Background technique [0002] In the process of human-computer interaction, the first action of the dialogue robot system is to obtain the target sentence input by the user. After the target sentence input is obtained, the server will analyze and identify the target sentence input by the user, so that it can A more appropriate answer to a user's question or request for feedback. [0003] In a dialogue robot, there are usually some intentions with more important meanings, such as consulting contact information, etc., and the operation of obtaining contact information is directly related to the success of the dialogue. However, the target sentence input by the user is more than one sentence, it may be many separate sentences, and it may also be sentences with uppe...

AI Technical Summary

Problems solved by technology

For example, the updating method and system, electronic equipment and storage medium of a text classification model disclosed in the Chinese invention patent CN111737472A need to collect the questioning operation performed by the user on the original classification result, and the webpage text and the webpage text The classification result of artificial mark is used as new sample; The data set that will comprise original training set and described new sample is divided into training set, test set and verification set, retrains text classification model; According to text classification model in training set, The F1 value on the test set and the verification set screens the best model, and updates the target text classification model to the best model, which consumes a lot of labor and time costs

Images