This process can be described as:
• Edition of CB messages, these messages have inserted not
visible options of reply in the first presentation screen of
the message, these options will be visible in case of the
user has interest and makes the first acceptance click.
• Messages CB are sent to the users located inside a specific geographic region (example: malls and restaurants),
increasing the possibility to reach a public with similar
preferences. These messages are sent through BSC/RNC
as indicated in Figure 2 (items 1 and 2).
• User, in case he/she has interest in the CB message,
chooses one of the reply options (second click) and he/she
answers by an SMS through the SMSC (Figure 2, item
C. Subscribers Segmentation
According to [11], in the personal recommendation systems
the user needs to be identified by the system to receive
recommendations. As shown in Figure 4, the user needs to
have a profile, because is behind this that the system can do
recommendations.
This paper will consider the personalized recommendation
systems with implicit profile, which is generated from collecting data from users’ responses to CB messages.
Fig. 4. Generic architecture of a recommendation system.
In the approach based on content, the recommended products or items are similar to those already purchased or
used by the client. In these systems should be modeled the
similarity notion between items. This similarity may simply be