Research On Multi-Service Oriented Network Intention Translation Method

With the rapid development of the Internet,network services emerge endlessly.The traditional network operation model has been unable to quickly provide users with a service experience that meets current needs.Intent-based networking(IBN)appears like the time required for rapid deployment.The research of IBN involves many aspects,among which intent translation refers to converting the intent into the corresponding network configuration requirements according to the user’s intent and the current network state.In recent research,most intent translation methods use specific templates to guide users to express their intent,and the application scenarios are relatively single,which cannot meet the needs of multiple business scenarios.As an essential part of IBN implementation,intent translation urgently needs a more practical,convenient,and widely applicable translation method.Based on the multi-service-oriented IBN scenario,this thesis studies the translation method of user network service intent,specifically,translating the user’s intent of different types of services expressed in natural language into routing configuration decisions.The main work of the thesis includes two parts:1.The thesis proposes a user intent translation method based on IBN and deep learning.Firstly,a joint intent classification and slot filling task model based on LC-BERT is proposed according to the characteristics of short text with intent in network business scenarios.Based on the original BERT model,the LDA and CRF models are integrated to solve the semantic problem to a certain extent.The problem of feature sparseness and lack of intent information are solved.It predicts labels more accurately.Secondly,the corresponding structured service routing configuration message is constructed,and the intent translation process is completed.Finally,the feasibility and effectiveness of the intent translation method in this thesis are verified by experiments.2.The thesis designs and implements a multi-service-oriented intent translation system,which realizes the transformation from the intent expressed in natural language to the routing planning strategy.The user enters the business intent statement through the web interface.The system generates the corresponding routing configuration message and routing path through the intent translation method and routing policy generation algorithm.It is verified that the system can normally work through the test.Compared with the existing technologies and research methods,the multi-service-oriented intent translation method proposed in this thesis supports users to express their own business needs in natural language.Users do not need to learn a specific policy language,which is more flexible and convenient.At the same time,it supports the translation of various network services,and the applicable scenarios are more abundant.