Decision Trees and Recommender Systems - Data Mining with Decision Trees: Theory and Applications - page 254

shared target attribute value or the most common value in case there is
more than one such value.
Decision

trees can be used for different recommender systems

approaches:

• Collaborative Filtering (CF) — Decision trees can be used for building a

CF system. Each instance in the training set refers to a single customer.
The training set attributes refer to the feedback provided by the customer
for each item in the system. In this case, a dedicated decision tree is
built for each item. For this purpose, the feedback provided for the
targeted item (for instance like/dislike) is considered to be the decision
that is needed to be predicted, while the feedback provided for all
other items is used as the input attributes (decision nodes). Figure 16.1
illustrates an example of such tree, when the item under discussion is
amovie.

• Content-Based Approach — Content features can be used to build a

decision tree. A separate decision tree is built for each user and is used
as a user profile. The features of each item are used to build a model that
explains the user's preferences. The information gain of every feature is
used as the splitting criteria. Figure 16.2 (right) illustrates Bob's profile.

The
Godfather

Dislike

Like

Pulp
Fiction

Star
Wars

Dislike Like

Dislike

Like

D

Fight
Club

L

The Sixth
Sense

Dislike Like Like Dislike Like

L D

L D

Fig. 16.1 A CF decision tree for whether users like the movie “The Usual Suspects”
based on their preferences to other movies such as The Godfather, Pulp Fiction etc.
A leaf labeled with “L” or “D” correspondingly indicates that the user likes/dislikes the
movie “The Usual Suspects”.