Visual Interactive Recommender
Recommendation Systems have gained a lot of importance in many different areas of application. They aim to replace the human adviser and thus help to deal with the current information overload. Therefore, such systems take a huge amount of information into consideration and mostly provide a personalized list of top n elements that might be of interest for the individual user. How the recommendations came about and which relationships have been utilized cannot be gathered from the outcome. As a result, the user experiences a lack of understanding of the process, under which the user's trust in the system ultimately suffers. One approach to address this issue are visualization techniques. With their help it is possible to visualize more complex results which do not omit too much valuable information.
One scenario, where visualization techniques together with recommendation systems are applied, is a publication network. This is a graph of information entities, like publications, and users, for example authors, with different kinds of relations, like co-authorship. Research has to focus on how the merge of a recommendation system and visualization can optimize the recommendation experience, for example by benefitting from the user's interaction with the system to improve the comprehension of and trust in the system.