Statistics visualization for the CERN Document Server

The CERN Document Server (CDS) is an open-source platform developed at CERN, that provides an online platform for storing and accessing multimedia content regarding High Energy Physics, such as documents, articles, images, videos and other digital scientific material. It is a thin software layer, built on top of Invenio, a digital library framework that enables the construction and management of large-scale digital repositories. Invenio is a highly modular framework, that consists of multiple  individual modules, each providing well-defined services and functionality to client applications.

As in most state-of-the-art applications hosted on the Web, in order to efficiently manage large-scale collections of digital records and simplify administration tasks, it is essential to periodically collect user-related statistics, extract any meaningful information from complex datasets and demonstrate their overview in a both comprehensive and appealing manner.

In this respect, the main objective of this project is to study techniques for visualizing statistics coming from the CERN Document Server, and more specifically design and implement a flexible Invenio module from scratch, needed for the upcoming version of the CDS. The Invenio-Stats-Js[1] module will undertake the task of building interactive graphs and visualizations of user statistics related to CDS records, providing useful feedback about traffic and popularity. The input data for the Invenio-Stats-Js module is retrieved from an already configured Elasticsearch cluster, currently running in production. The integration of such module into the production system of CDS can significantly boost the productivity of developers and also enhance the QoS offered to end-users.

Most of our work was targeted for the new version of CDS Videos (v3.0), which was a top priority during the migration phase. However, since the functionality provided by the Invenio-Stats-Js module is widely applicable in different software environments, it was decided to package it and release it as an independent module on NPM, to facilitate its public distribution, continuous integration and future maintenance.