Poster: Word embedding for deployment descriptors in NFV

Virtualized networks such as NFV enable the automation of the network and decrease the need for human interventions. They ensure the availability of adequate compute, storage, and network resources for network services. To automate the network, the NFV orchestrator coordinates and manages automatically the resources to meet the requirement of network functions and services. To date, network function deployment descriptors are manually designed by service providers. Learning automatically from deployment descriptors is an important step toward fully automating the network management. In this work, we propose a word embedding approach that learns from a set of deployment descriptors a vector representation. This approach is shown to be useful for recommending deployment descriptors.