Page 236 - Big Data Analytics for Intelligent Healthcare Management

9.2 BACKGROUND 229

obtains user interest indirectly through monitoring user behavior. The combination of both explicit and
implicit feedback is known as hybrid feedback [14, 15].

• Learning phase
This phase takes the feedback collected in the previous phase as input and processes this feedback by
using a learning algorithm and exploits the user’s features as output [14, 15] (Fig. 9.1).
• Prediction/Recommendation phase

Preferable items are recommended for users in this phase. By analyzing feedback collected in the in-
formation collection phase, predictions can be made through model- or memory-based or observed
activities of users by the system [14, 15].

9.2.3 METHODOLOGY
9.2.3.1 Filtering techniques
An efficient recommendation technique is necessary to provide useful recommendations to end users.
This section describes three recommendation techniques that are mainly used for providing recommen-
dations to users about products. The following figure shows the hierarchy of recommendation systems
based on different filtering techniques [15] (Fig. 9.2).
A. Content-based filtering technique
The content-based filtering technique focuses on the analysis of features and attributes of products to
generate predictions. Content-based filtering is usually utilized in document recommendation. In this

Information
collection phase

Learning phase
Feedback

Prediction/
recommendation
phase

FIG. 9.1
Phases of the recommendation system.