Digitised dutch cultural heritage, Mahout & Hadoop
In 2009 I was asked by the Dutch Institute of Sound and Vision to implement a concept that resided in a national project called ‘Images for the future’, with the goal to save and make available important Dutch audio-visual heritage collections of the 20th century (see: https://imagesforthefuture.com/en/project).
Within this project different kinds of software-frameworks are produced, one of them being a Recommender Engine, which provides users of content-platforms (semantic)-recommendations based on their user-profile.
In the early stages of exploring available, mature and open frameworks we stumbled upon Taste and Mahout. After more research we built functional specifications on top of Mahout and asked some IT-providers in Amsterdam to come up with an implementation plan. As we speak, we are about to release the project ZieOok (AlsoSee), which basically runs on Mahout, but has an administrator-Dashboard, enabling content-owners to create specific recommenders based on their preferences (time, specific part of collection, what kind of algorithms to be used etc.). ZieOok therefore has its own Dashboard and a Ruby REST API that communicates with Mahout.
As of April, two dutch archives will interface with ZieOok, one of them being www.uitzendinggemist.nl, the largest VOD platform provided by the NPO (Dutch National Broadcaster) in the Netherlands.
The abstract of the talk will be to present the audience with a Lightning format (10min) to briefly introduce the history of this project (why and how) and to demonstrate how content-platforms can easily enter this framework and how it could potentially benefit that specific platform.
