Design And Implementation Of Intelligent External Call Platform System

With the improvement of living standards and the development of the Internet,people can complete the purchase,reservation and other things without leaving home,while the important way for enterprises to sale,pay a return visit and other tasks is to call out,which is also an important means to expand the market and maintain customers.The traditional manual outbound mode has many problems and difficulties in personnel training,operation cost,user experience,precision of data recording and other aspects.With the development of artificial intelligence,the use of robot integrated system for external call has attracted extensive attention from enterprises and researchers.According to the above problem,intelligent call platform system based on speech technology,use of speech recognition,speech synthesis,voiceprint recognition,natural language understanding technology,intelligent human-computer interaction technology,mainly to provide users with intelligent voice interaction service and various functional strategy configuration services,applicable to the products or services to promote demand.The basic function modules of this system include permission management module,customer management module,external call task management module,call management module,robot configuration module,man-machine dialogue module.These modules are also the modules I am mainly responsible for.Among the man-machine dialogue modules,the main functional modules described in this paper include the NLU intention recognition function,dialogue management and natural language generation applied to task-based multi-round dialogues.The intention recognition function in the system adopts the mode of combining traditional machine learning method and deep learning method.The method of dialog management is based on finite state machine,using directed graph structure and adopting the strategy of gradually confirming each information,showing the business process and state that should be defined in the dialog system.Natural language generation adopts the mode of natural language generation based on template which expresses efficiently.The front-end website of this project uses the vue.js framework,and the back-end adopts the overall micro-service architecture pattern.Spring Cloud is used as the main framework,and springboot+eureka+zuul+hystrix and other components are used.Grpc remote procedure call method is used for communication between services,and Rabbit Mq messaging middleware is used for asynchronous communication and message push due to the change of call state.The database is combined with Mysql and Redis,and the log system uses kafak for storage.At present,this project has been completed and launched,and a million data volume has been called out,which provides great convenience for enterprise customers to complete the assigned tasks and obtain user information.