Recommender systems have been strongly researched within the last decade. They have strongly progressed over the time for non-live evironments. However, they do not meet the requirements of live broadcast. Live broadcast suffers from several inherent problems, e.g. the impossibility to foresee the contents of the next items or the reactions of the user to the changing programme.
In this thesis a simple model of the complex live broadcast environment is developed to classify the media assets and the reactions of the viewer. A content based and a hybrid recommendation algorithm are tested. The simple but powerful content based algorithm has proven to calculate high quality recommendations from low quality data. The hybrid approach does not perform that well. However, this is caused by the poor base data.
The thesis can serve as a basis for further live recommender system research.
recommender systems, interactive TV, live TV, sports transmission
@mastersthesis{Gude:2007, author = "Gude, Martin", title = "Recommender Systems for Live-Transmission in interactive Television: Gathering and Rating User Feedback", keywords = "recommender systems, interactive TV, live TV, sports transmission", type = "Master's Thesis", year = "2007", address = "Cologne", school = "University of Applied Sciences Cologne", collaboration = "Gr{\"u}nvogel, Stefan M. Prof. Dr. and Pla{\ss}mann, Gerhard Prof. Dr.", url = "http://martin-gude.de/go/master"}
@inproceedings{GudeGP:2008, author = "Gude, Martin; Gr{\"u}nvogel, Stefan M.; P{\"u}tz, Andreas", title = "Predicting Future User Behaviour in Interactive Live TV", keywords = "recommender system, interactive live TV, multistream", booktitle = "Changing Television Environments, 6th European Conference, EuroITV 2008, Salzburg, Austria, July 3-4, 2008, Proceedings", year = "2008", editor = "Tscheligi, Manfred; Obrist, Marianna; Lugmayr, Arthur ", pages = "117-121" publisher = "Springer", series = "Lecture Notes in Computer Science", url = "http://martin-gude.de/go/master", volume = "5066", date = "2008-08-12", ee = "http://dx.doi.org/10.1007/978-3-540-69478-6_14", isbn = "978-3-540-69477-9", }
The thesis is licensed under a Creative Commons Licence. If you want to use it under another licence feel free to contact me.
The short paper was presented at EuroITV 2008. It was published in the conference's proceedings and is thus protected by copyright (© Springer-Verlag Berlin Heidelberg 2008).