Jupyter Notebook for Constructivist Digital Learning
Project members: Sebastian Klaßmann, Timo Varelmann, Nils Dahmen, Uwe Seifert
In 2018, we implemented an interactive coding environment within the University of Cologne’s intranet that is to be used to support teaching and learning processes in the disciplines of computational and cognitive musicology after having successfully tested it on an external test server earlier this year - the Jupyter notebook. This platform is based on the cell-based and interactive Jupyter Notebook (JSON) format and is served using Jupyterhub on a Virtual Machine hosted by the RRZK. The main applications of Jupyter Notebook are interactive text editing and execution of code written in Julia, Python or R. A given notebook is a combination of cell-based, active code elements that are expanded upon by advanced text editing features (using Markdown and LaTex). We have been using Jupyterlab as our graphic frontend (GUI), because it is constantly improved upon by the Jupyter Project. Furthermore, openly available extensions can be used to easily expand Jupyterlab’s core functionality - for example, by enabling the use of LaTex code, adding converter functions for established file formats (such as PDF), or by providing the option to convert any given notebook into a multi-dimensional presentation in the style of reveal.js.
The use of Jupyter Notebooks provides teachers and students with an explorative and exemplary approach to fundamental methods used and topics discussed in the realms of computational and cognitive musicology. Jupyterhub can be purposefully used in seminars in order to extend the physical and social interactive spaces of traditional teaching environments by adding to it a digital space and thus constituting a hybrid, third space or to offer pure online courses. Technically, this is achieved by providing individual, containerized workspaces that can be accessed from any internet-capable device using a contemporary browser. This almost entirely eliminates the potential for hardware and software incompatibilities, which have frequently occured in the past when using single-user local setups. Thus, Jupyterhub enables the application of blended learning principles by combining traditional course structures with E-Learning options, such as online course materials, data structures for musical analysis and virtual classrooms with individualized assignments.
A study group has started to evaluate traditional academic workflows based on our first experiences with using Jupyterhub in a university teaching context. A second focus of this study group has been the reflection of current, interdisciplinary research approaches in the field of computational musicology. Furthermore, various seminars have enabled us to try out the possibilities of adaptive learning formats and individualized assignments based on our architecture, as well as to evaluate the didactic implications of working with Jupyter Notebook. Due to these processes and first seminar papers that have been conceptualized, written and submitted entirely within our hub architecture, we have collected first-hand empiric data that warrants a well-founded development of this system towards a concise concept for digital learning and teaching. Based on the discussion in (Rossant 2018), we are already very close to the state-of-the-art of interactive computational teaching and learning.
Currently, we are trying out version control software (GIT) and it’s extensions (GitHub, GitPages) as a means to digital academic publication and decentralized workflows. They have been successfully integrated in various seminars. The website for our seminar computational musicology (winter term 2018/2019) might give a good impression of the possibilities this offers (https://sebastianklassmann.github.io).
In the near future, we want to expand our framework by implementing technical possibilities for real-time digital collaboration, as well as building and expanding upon our very own corpus of relevant, exemplary models and experiments that are to be represented as Jupyter Notebooks. Furthermore, we are planning to test the possibilities offered by pure online seminars based on Adobe Connect and Basic Support for Cooperative Work.
Additional Literature:
Rossant, C. (2018). IPython Interactive Computing and Visualization Cookbook: Over 100 hands-on recipes to sharpen your skills in high-performance numerical computing and data science in the Jupyter Notebook. Packt Publishing Ltd.