NHTV University of Applied Sciences - Academy of Digital Entertainment The intelligent Media Intermediary (IMI)
The latest generation of mobile phones are called smart phones. But how smart are they really? Do they know you? Do they know what you want and when and where you want it? In principle it should: the devices contain information on your calendar, your location, your surfing behavior, your interaction with friends in social media, your bookmarks, the apps you use and even bank account en payment information. But still it doesn’t utilize the full potential of this rich source of data. The phone knows you, but it doesn’t understand you.
The Intelligent Media Intermediary project has the objective to improve the smartness of smart phones, by building a platform that delivers recommendations to its user at the right place, at the right time. In doing so the platform delivers new possibilities for commercial parties as well: the aim is to build a device that targets offerings in the most effective and at the same time non-intrusive way: at the moment the user is interested. The IMI platform will be a platform open to third parties: they can develop apps that make use of the intelligence of this platform.
One of the starting points in the development is to solve a second issue, which is privacy. The IMI platform will be build in a way that, others than currently the fact in many new media applications, the consumer is the owner of its data: all recommendations are based on permission and the user can always decide to cut the line between its phone and the database.
NHTV has chosen for a hybrid approach: software engineering (rapid prototyping) and research take place simultaneously, thus creating a mechanism that reinforces itself: research improves the platform and the platform is being used to collect user data which are needed for further research.
The research part of the project focuses on two issues: recommender systems and activity recognition. Recommender systems are known for years: web shops like Amazon and iTunes provide customers with recognitions like” people that bought this. Etc…” in most cases these recommendation are limited to one platform. IMI aims to extend the use of recommendations to a broader context, using a combination of collaborative filtering, content based recommendations and knowledge based recommendation systems.
A second research goal is to confirm that user context influences the way the user perceives the relevance of information. Pushing information to the user is extremely relevant, but as our research shows, it is more effective in the right context. Pushing information without being aware of the user context, impacts the relevance and acceptance of the information. For example, it is not relevant to know that it is starting to rain in the middle of the night while you are sleeping; it is more relevant just before you leave your house in the morning. Or you probably want to know that there is a traffic jam in the highway and it would be best if you would take the train today, but this suggestion is not so much useful, when you are already stuck in traffic.
Therefore, this is function specification of a middleware platform, which allows 3rd party entities to recommend activities to users considering their context. IMI is being developed to be a support tool to the users, but at the same time to conduct research by evaluating user’s preferences and improve the profiling analysis.
The software engineering part of the projects will be rolled out in steps. Currently we are building the platform and the data collection application. The first application will be a student app, developed together with commercial external parties. It will deliver students information related with their study (like schedules, marks), it will connect this information to third party information (like public transport information) and will be open to third parties (like restaurants, cinema, pubs), thus creating a system that makes the life of students easier, that at the same time serves as a data collection tool for researchers. A second step will be to develop a TV app, that serves as a second screen and enhances the data with media behavior.