Abstract :
Technological developments in the last few years developed rapidly, especially in
information technology. This development brought a serious impact on the number
of data and the spread of information on the internet so required the existence of a
recommendation system to sort the information in accordance with the needs of the
user. This research method using Mixed Hybrid by combining the two methods that
is Content-based Filtering and item-based Collaborative Filtering. The merger of
both of these methods are aimed at overcoming the weaknesses of each method.
The algorithms used, namely Apriori on a Content-based Filtering and Cosine
Similarity Centered on Collaborative Filtering. The experiment was performed on
FilmKu information system by using Movielens dataset-100k. The results obtained
in the form of a list of 10 recommendations of movies with a value of MAE and the
ratio of the largest 0.8779 in ruleset reached 42%.
Keywords: Apriori, Content-based Filtering, Centered Cosine Similarity, Itembased
Collaborative Filtering, Lift Ratio, MAE, Mixed Hybrid, Movielens,
Recommendation System.