Font Size: a A A

The Design And Implementation Of XIAO AI Intelligent Customer Service System Of Htsc Platform Department

Posted on:2021-05-18Degree:MasterType:Thesis
Country:ChinaCandidate:D K LiuFull Text:PDF
GTID:2428330647450847Subject:Engineering
Abstract/Summary:PDF Full Text Request
When traditional manual customer service is facing a variety of problems from different businesses,it often repeats a lot of mechanical and boring work,which brings high labor costs to the enterprise.In this context,it makes sense to design and develop an intelligent customer service system to improve work efficiency.Based on the questions raised by customers,the system provides answers in the corresponding fields,which can replace manual customer service in a certain sense,greatly improving the content of financial technology and saving labor costs for enterprises.The solution of the intelligent customer service system for answer-querying includes two parts: intent recognition and answer generation,in which the intent recognition uses few-shot classification and short text classification,and the answer generation uses the knowledge base to match answer.Few-shot classification is implemented based on the Matching Network,and short text classification is implemented based on the LSTM Network.The back end of the entire system is developed using the currently popular Spring boot framework and the system is deployed as a web application.The algorithm module of the system is developed using Pytorch.This intelligent customer service system supports customers to query answers and submit feedbacks,supports managers to manage knowledge base data such as questions,intents,and answers,and supports managers to train and predict the intent recognition models.It basically completes the main functions that a customer service system should have.Through this system,some customer service tasks related to business can be completed without manual work.The solution used in this system has the advantage of flexibility.Due to the separation of intent recognition and answer generation,different enterprises can customize the intent area according to their own circumstances,and can also customize their own knowledge base.So,it can be switched according to different businesses and different areas.This system has been deployed and operated in Huatai Securities.
Keywords/Search Tags:Intelligent customer service system, Intent recognition, few-shot classification, short text classification
PDF Full Text Request
Related items