Service traffic contains two aspects: score and form. The score
implies the degree of interest a user has in multimedia traffic,
while the form denotes the content features on a particular
device. For the purpose of efficiently operating various types
of multimedia traffic, we classify the data into three categories:
preference data, situation data, and capability data. At
first, the multimedia traffic server calculates the similarity
between the media service and the preference data by employing
the method in [7]. Then it evaluates the probability of the
media service belonging to one of the servers. We can get the
score from the weighted sum of the above calculated similarity
and probability.