Decision Trees and Recommender Systems - Data Mining with Decision Trees: Theory and Applications - page 255

Genre is
Comedy

No

Ye s

Staring by George
Clooney

Published
year

Ye s

No >2000

<=2000

D D

L L

Fig. 16.2 A CB decision tree for Bob.

It should be noted that although this approach is interesting from a
theoretical perspective, the precision rate that was reported for this
system is not as accurate as that of recommending the average rating.

• Hybrid Approach — A hybrid decision tree can also be constructed.

Only a single tree is constructed in this approach. The tree is similar to
the collaborative approach in the way it treats the user's attributes as
attributes to split by (such as her liking/disliking of a certain movie).
But the attributes the tree utilizes are merely general attributes that
represent the user's preference for the general case and that are based
on the content of the items. The attributes are constructed based on the
user's past ratings and the content of the items. For example, a user
that rated negatively all movies of the comedy genre is assigned a low
value in a “degree of liking comedy movies” attribute. Similarly to the
collaborative approach, the constructed tree is applicable to all users.
However, it is now also applicable to all items since the new attributes
represent the user's preferences for all items and not only for a single
given item. Figure 16.3 illustrates such a hybrid tree.

Consider a general case with a dataset containing n users, m items,
and an average decision tree of height h . The collaborative filtering based
RS requires m trees to be constructed, one for each item. When a user
likes to receive a recommendation on what movie to watch, the system
traverses all trees, from root to leaf, until it finds an item the user would
like to view. The time complexity in this case is therefore O ( h m ). This

mightbetooslowforalargesystemthatneedstoprovidefastand

·