This workshop has been cancelled due to too little participants
Description
The interest in recommender systems has increased immensely over the past years, and recommenders have been showing up in various online services. In addition, more and more research is being done on the subject, and there is now even a yearly conference fully dedicated to recommender systems. However, both researchers who want to experiment with recommender systems and companies that want to use recommender systems in their applications have to build their own recommender systems from scratch, thus struggling with implementation issues that others have already solved.
There are a few companies that offer commercial recommender system solutions. Unfortunately with their products it is often difficult to determine what these recommender systems actually consist of, what kind of algorithms are being used, and how if any, algorithms are being combined. Whereas extending such software is often difficult, changing the software for research purposes is even more difficult, if at all possible, and the results cannot be shared with others in the research community and industry.
The focus of this workshop is setting up a software project to develop open source recommender system software and creating a community around it. Such software can be used by researchers as a starting point in their research that can easily be extended with their own ideas. Sharing the implementation of their ideas in an open source project will give them more exposure to a wider audience. Companies creating or using recommender systems on the other hand get a head start from using open source recommender system software in their own applications, and benefit from sharing their own software by a decreased need to keep their work in sync with a changing open source base.
There are a few open source recommender system projects. Most focus on only one recommendation algorithm (e.g. TASTE project which only focuses on collaborative filtering). Others have been set up, with no more information than that they intend to develop and release recommender system software. Some of the organizers of this workshop have released an open source recommender system framework called The Duine Framework, which contains a set of Java software libraries to create recommender systems for applications that can use multiple recommendation algorithms. The Duine framework is far from finished. It currently only contains a few basic recommendation algorithms and one evaluation method. There are many more issues that need to be addressed. Another such project is The MyMedia project which develops a .NET framework with advanced recommendation algorithms that will be open sourced, at least for research purposes. Some of the organizers of this workshop are also involved in the MyMedia project.
Furthermore, the development of open source software is best done by a community of researchers and developers from various organizations. That way, more resources can be put in the development of the open source recommender system software and a community is also a better guarantee for the long term continuation of such a project.
Therefore, the workshop has two main goals:
This workshop has been cancelled due to too little participants
