Design And Implementation Of Intelligent Outbound System

With the development of Internet technology and artificial intelligence,various industries are also developing rapidly.In this process,communication between the company and the customer is indispensable,such as product promotion,business guidance,and debt collection.The traditional way of communication between enterprise and customer is often the call center automatically calling out,and then transfer to the artificial agent to complete the outbound call task.This kind of manual-based customer communication is accompanied by a large number of outbound calls,dull work,low call efficiency,and irregular speech.Aiming at the above problems,an intelligent machine outbound system is constructed based on the traditional call center in which the semantic parsing engine including intent recognition and intelligent question answering function and the state machine based dialogue management are designed and implemented.Artificial intelligence based outbound call system can effectively solve the problem of high cost and low efficiency of manual outbound call in the traditional outbound call process,so it has great practical and commercial value:(1)Text classification based intent recognition:Firstly,the research status of natural language understanding technology at domestic and abroad is investigated.According to the actual characteristics of the intelligent outbound task,the intent recognition based on text classification technology is realized.A deep learning intent recognition model based on word,word and pinyin vector features is designed,and the intent recognition method of rule template is used to solve the problem of lack of data during cold start.(2)Depth semantic matching based question answer:In actual scenarios,user problems are often not exhaustive,and as the amount of users accumulates,the set of problems will continue to expand,so a certain number of intent recognition methods cannot satisfy the scenari o.The deep semantic matching model is used to judge the similarity between the user problem and the standard frequently asked question,and then select the appropriate speech feedback to the user.Experiments show that this scheme can significantly improve the user experience and actual effect of the intelligent outbound system.(3)The design and implementation of task-oriented dialogue system for outbound:Based on the above key technologies,a complete intelligent machine outbound system is designed and implemented.In the actual business scenario,it is verified that the system can complete the scheduled outbound task and easily migrate to other business scenarios.