Forzify - a SPORTIFY system for sport videos
To target and engage supporters and fans, new ways to experience for example soccer are needed. Especially young people do not necessarily “just” watch a game in the stadium or on TV at home. They are simultaneously interacting with friends on social media. They want to share their current experience with them, and expand this experience into a broader story by searching for similar events on YouTube and generate own content using their mobile phones. These new interaction patterns must be taken into account in next generation systems for both in-stadium experiences and external live and on-demand services.
Description of topic
What we aim to research in this thesis is the next generation video experiences for soccer by transforming how soccer fans and supporters interact through novel technological solutions – not only on the day of the match, but 24/7. The idea is to develop a Spotifly, called Forzify, for soccer events where the system gives the users an interactive social experience. The idea is to boost supporter activity where the fans are able to see highlights and summaries of games, collect and redistribute favorite soccer clips, compose their own summaries of soccer events, share and discuss them with others, and get recommended by fellow fans, all in a social experience. Thus, search for your favorite game happenings, see video compositions made by your friends, and let the system present and recommend the most played and “liked” events from the previous week – in the order you want …
A first version is already deployed for Vålerenga IF (an elite soccer team in Norway), see vif.forzasys.com or download the app "Forzify VIF" from app-store (and soon google play) to see a running example. However, as you can see, there are many aspects that can be improved...
1. search results and recommendations
2. investigate usability in other sports
3. scaling to millions of Real Madrid fans
The company ForzaSys has earlier developed a first prototype of this kind of a system. The idea is built on the de facto standard video delivery solution today using adaptive segmented HTTP streaming. The system uses existing metadata to generate (searchable) events, and playlists of such events are generated by manipulating the manifest files used to download the video segments. A preview of such generated video summaries may be shared freely on social media. The search patterns and the social interaction are logged and analyzed in order to give better search results and better recommendations. There are several possible topics in the context of this project, but the main topic is to look at the search and recommendation system in the back-end possibly running in the cloud using some kind of distributed processing framework like Apache Hadoop (MapReduce) or Microsoft Dryad.
- Pål Halvorsen
- Carsten Griwodz