The results found, such as the duration time distribution
of online videos and its number of tags, are important for
learning about our database, and it can contribute to the
decision made during the entire research.
We also propose a recommendation technique based on
the object being consumed. This method has the goal of
the generation of similarities rankings between the database
items (in most cases, online videos). From this, we elaborate
three models that uses different dimensions to compare
items, an compare it with a state-of-the-art technique called
WRMF. The experiments exposed demonstrate that, the
more information considered about the video consuming is
used for the ranking generation, the better are the recommendations
results. Experimental results indicate that the
proposed method is very promissing, which had obtained
almost 70% in precision. We also perform distinct evaluations
using different approaches from literature, such as the
state-of-the-art technique for item recommendation.
The results presented in this paper are important for both
user and provider of Multimedia Web Content, and it can be
applied in the generation of personalized services, besides
the recommendation of this content. As future work, we are
going to explore the information about the user, besides the
utilization of different dimension for similarity calculation
between items. Finally, in partnership with Sambatech, we
are going to apply the techniques developed in this paper in
a real environment to improve its recommendation service.
VIII. ACKNOWLEDGMENTS
This work was partially sponsored by Sambatech
(www.sambatech.com.br) and partially supported by the
Brazilian National Institute of Science and Technology for
the Web (CNPq grant no. 573871/2008-6), CAPES, CNPq,
Finep and Fapemig.