Human-computer interaction dialogue management system

Abstract

The invention provides a man-machine interaction dialogue management system. The dialogue management system comprises a dialogue system and a data system, the dialogue system is used for identifying user intentions, the data system records external use conditions for the dialogue system and enables knowledge for the dialogue system, and the data system comprises a corpus management module, a script generation module, a test execution module and a data management module. Corpus information of the dialogue system is stored through the corpus management module, then corresponding scripts are generated from corpora of the corpus management module through the script generation module, the scripts are loaded into the dialogue system, and knowledge enabling is conducted on the dialogue system; verifying the identification condition of the modified script through a test execution module; the data management module is used for collecting dialogue records of users, quickly screening out trainingcorpora and adding the training corpora into the corpus management module, and meanwhile, generating a hit rate statistical graph to facilitate adaptive adjustment of products; according to the invention, the recognition rate of human-computer interaction can be improved.

Description

technical field [0001] The invention relates to the technical field of artificial intelligence, in particular to a dialog management system for human-computer interaction. Background technique [0002] At present, the common intelligent dialog frameworks are roughly divided into two categories. One is a rule-based framework, such as aiml. Developers can write corresponding aiml scripts, and then load them into the dialogue system for training; the other is a learning framework, such as: Rasa , developers can achieve a dialogue intent recognition by writing corresponding corpus scripts for training. AIML, full name Artificial Intelligence Markup Language (Artificial Intelligence Markup Language), is an XML language for creating natural language software agents. Rasa is an open-source machine learning framework for automated text and speech-based conversations. [0003] Disadvantage 1. For rule-based frameworks (such as aiml), in order to support various questions that robot...

AI Technical Summary

Problems solved by technology

[0003] Disadvantage 1. For rule-based frameworks (such as aiml), in order to support various questions that robots can recognize intent, it is necessary to artificially expand the corpus, and then write it into a corresponding script to ensure that more questions can be recognized (such as : Configure a sentence "#How#improve#productivity#", then enter "how to improve productivity", aiml cannot recognize, you need to configure a sentence "#how#improve#productivity#")

[0004] Disadvantage 2. For learning frameworks (such as rasa), it supports generalization capabilities and can identify relatively similar questions (for example, the above-mentioned configuration "how to improve productivity" and asking "how to improve productivity" can be recognized), But the error rate is relatively high, especially on some sensitive issues, the generalization ability may make the robot answer irrelevant questions

At the same time, the training framework also needs to write scripts. Compared with the regular framework, the maintenance cost is low, but the recognition error is relatively high

[0005] Disadvantage 3. No matter which framework it is, it currently requires the cost of human maintenance. It is necessary to write corresponding scripts to improve the "brain" of the robot, so as to endow the brain with the ability of knowledge points

There may be duplicate data in the collected user data, which will lead to duplicate configuration and waste communication costs

[0006] Disadvantage 4. Most of the dialogue system tests are just to check whether the expected results of the known problems are correct.

There is no record of the dialogue growth of the robot, and it is impossible to know the experience and development status of the current dialogue system in the actual application process

Images