Design And Implementation Of Dialogue Configuration Rule Mechanism For Smart Home System

The increasing popularity of cloud computing technology and the maturity of artificial intelligence technology are driving the arrival of the Internet of Things era.Smart home can be regarded as an important application of the Internet of Things,which aims to provide users with a"people-oriented" new home life experience.In the smart home system,the system provides services to users by executing rules preset by users.Therefore,how to let users complete the configuration of preset rules simply and efficiently is very important for users.Based on the above considerations,this topic proposes a dialogue configuration rule mechanism for smart home systems.After applying this mechanism in the smart home system,users only need to have one or more rounds of dialogue with the system to complete the configuration of the rules.The application of this mechanism is of great significance for improving the user experience and even for the promotion of smart home systems.To achieve this mechanism,the main work of this paper is as follows:First,in terms of system modules,this paper analyzes functions of this system and according to these functions divides the system into the following main modules,including voice information processing module,natural language understanding module,dialog management module,natural language generation module,JavaScript code generation module,and data transmission module.Based on the Flask framework,the development of each module is achieved.Secondly,in terms of algorithm research,this paper uses a recurrent neural network RNN to train the data in the corpus to obtain a suitable slot filling model,and applies this slot filling model to complete the task of slot filling.In addition,in order to ensure the accuracy of the slot filling results,this paper also uses the dependency syntax analysis algorithm to mark the dependency relationship of the user input text,and further validates the slot filling results through the dependency relationships.Finally,with regard to the data set,I have participated in the development of the IoT cloud platform system of our laboratory which is currently online.Based on the actual situation,the essence of the platform at this stage is a smart home system.This paper obtains the smart home rules configured during the actual use of the platform as part of the data set.However,since the number of existing rules in the existing smart home system is not enough to train a more suitable slot filling model,this article combines the relevant device information of the platform to generate another part of the rule data set on its own.