|
Name |
Bruno Miguel Delindro Veloso |
|
Affiliation |
Universidad de Vigo |
|
Corresponding WG |
WG4 |
|
Grant research topic |
Big Data Techniques for Media Recommendation and Viewer Profiling |
|
Hosting Institution |
Birmingham City University |
|
Period |
2016-03-01 to 2016-03-31 |
|
Summary of the scientific report |
Nowadays, not only the number of multimedia resources available is increasing rapidly (according to BigMM (2015) it will become the biggest short term Big Data domain), but also the feedback information provided by the viewers, generates a huge quantity of information com- posed of ratings, likes, shares and posts/reviews. The large amount of multimedia resources produced every day surpasses the human filtering capability to search and find in real time media resources of interest. Recommender systems, which match users with resources, can be developed to address this problem. The user generated data, which corresponds to explicit user preferences and intrinsic behaviours, can be used for defining the user profiles. In particular, dynamic user profiling, i.e. the ability to build and update profiles based on the continuous stream of user interactions (likes, posts, ratings, watched items, etc.), can be addressed as stream mining João Gama (2010). Such an approach will allow us to evolve from the standard static off-line profiling to the dynamic on-line profiling. The work accomplished during the grant period explored, applied and assessed Machine Learning, Data Mining and Stream Mining techniques to build dynamic user profiles and make recommendations. |
|
Personal notes |
Big Data presents several data processing challenges as well as opportunities for the researchers to explore, investigate and exchange knowledge through international research programmes. This Short-Term Scientific Mission (STSM) grant provided an opportunity to learn and explore Machine Learning, Data Mining and Stream Mining techniques to apply on my PhD research domain. |
