Abstract
Cloud-network integration is a profound change in infrastructure brought about by the deep integration of communication and information technology. In this new environment, increasing experts begun to pay attention to the diversity and high dynamics of user intents and the variety of services. To enable users’ intent in different scenarios to be accurately captured and transferred to the lower-level module for Quality of Service (QoS) control, this paper proposes a method combining Bi-directional Long Short-Term Memory-Conditional Random Field based entity recognition technique (BiLSTM-CRF) and QoS accurate mapping technique, which can capture and translate users’ intent in cloud-network integration environments. This method takes into account the complexity of the intent in the new environment and can more accurately map the QoS parameters that users actually need (such as throughput, response time, etc.). Especially for ambiguous intents, parameters that match user’s expectations will be obtained.