As they are mainly based on bodily experiences and embodied knowledge, dance and movement practices present a great diversity and complexity across genre and context. Thus, developing a conceptual framework for archiving, managing, curating and analysing movement data, in order to develop reusable datasets and algorithms for a variety of purposes, remains a challenge. In this work, based on relevant literature on movement representation and existing systems such as Laban Movement Analysis, as well as working with dance experts through workshops, focus groups, and interviews, we propose a conceptual framework for creating, and analysing dance learning content. The conceptual framework, has been developed within an interdisciplinary project, that brings together technology and human computer interaction researchers, computer science engineers, motion capture experts from industry and academia, as well as dance experts with background on four different dance genres: contemporary, ballet, Greek folk, and flamenco. The framework has been applied: a) as a guidance to systematically create a movement library with multimodal recordings for dance education, including four different dance genres, b) as the basis for developing controlled vocabularies of dance for manual and automated annotation, and c) as the conceptual framework to define the requirements for similarity search and feature extraction.